Evaluation of Low Frequency Communication Techniques  
for Load Phase Identification 
 
by 
 
Daniel Lee Geiger II 
 
 
 
 
A thesis submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the 
requirements for the Degree of 
Master of Science 
 
Auburn, Alabama 
December 12, 2011 
 
 
 
 
Keywords: Power Line Communication Techniques,  
Low Frequency Communication, Frequency Shift Keying 
 
 
Copyright 2011 by Daniel Lee Geiger II 
 
 
Approved by 
 
 S. Mark Halpin, Chair, Professor of Electrical and Computer Engineering  
R. Mark Nelms, Professor and Chair of Electrical and Computer Engineering  
Charles A. Gross, Professor Emeritus of Electrical and Computer Engineering  
 
 
 
 
 
 
ii 
 
  
Abstract 
 
 
 This is a discussion on the process of implementing a system for power line 
communication (PLC). A PLC system is desired as it will allow for phase identification 
of the load and the transmission of data from the load to a central receiver. There are two 
main processes in a PLC system: a communications protocol and signal coupling. The 
communications protocol involves the methodology employed for successful 
communication between the load and the receiver. Signal coupling involves coupling the 
communication signal onto the power line at the load and off of the power line at the 
receiver. A successful implementation of these two processes will establish a system for 
PLC. The methodology employed for communication between devices is frequency shift 
keying (FSK). FSK will allow for low frequency operation with minimal attenuation. 
There are various coupling strategies that are presented for discussion. It is found that 
magnetically coupling a communication signal should be avoided and that a directly 
coupled signal generator should be implemented. 
  
 iii
 
Acknowledgments 
 
 
?For by him all things were created, in heaven and on earth, visible and invisible, whether 
thrones or dominions or rulers or authorities?all things were created through him and for 
him.?   
Colossians 1:16 - ESV 
  
 iv
 
Table of Contents 
 
 
Abstract ............................................................................................................................... ii 
Acknowledgments.............................................................................................................. iii 
List of Tables ..................................................................................................................... vi 
List of Figures ................................................................................................................... vii 
List of Abbreviations ......................................................................................................... ix 
Chapter 1: Introduction ........................................................................................................1 
Chapter 2: Communications ................................................................................................4 
 2.1: Slave Device Wait and then Transmit ..............................................................8 
 2.2: Master Asks and Slave Responds ...................................................................10 
Chapter 3: Coupling Introduction ......................................................................................11 
 3.1: Signal Coupling ..............................................................................................11 
 3.2: Data Signal Strength for Successful Recovery ...............................................11 
 3.3: Assumptions....................................................................................................13 
Chapter 4: Coupling Techniques .......................................................................................15 
 4.1: Series Current Injection Technique ................................................................15 
 4.2: Shunt Current Injection Technique .................................................................24 
 4.3: Series Voltage Injection Technique ................................................................30 
 4.4: Shunt Voltage Injection Technique ................................................................34 
 4.5: Switched Impedance Method ..........................................................................39 
 v
Chapter 5: Summary ..........................................................................................................43 
Chapter 6: Recommendations and Future Work ................................................................45 
References ..........................................................................................................................46
 vi
 
List of Tables 
 
 
Table 1: Example Data Log Based on Figure 1 ...................................................................3 
 
 
  
 vii
 
List of Figures 
 
 
Figure 1: Load Phase Connection Diagram .........................................................................3 
Figure 2: Positive Sequence Power Line PI Model .............................................................5 
Figure 3: Frequency Response of the Power Line PI Model ...............................................6 
Figure 4: Frequency Response of the Power Line PI Model  
with Adjusted Y-Axis Limit ................................................................................6 
Figure 5: Data Signal ...........................................................................................................9 
Figure 6: Example of FSK .................................................................................................10 
Figure 7: 130 Hz Signal Recovered by FFT ......................................................................13 
Figure 8: Series Current Injection Technique ....................................................................15 
Figure 9: B-H Curve of a W-type Material ........................................................................18 
Figure 10: Line Diagram of the Shunt Current Injection Technique .................................20 
Figure 11: Th?venin Equivalent Line Diagram from Substation to Load .........................21 
Figure 12: Shunt Current Injection Technique ..................................................................24 
Figure 13: Line Diagram of the Shunt Current Injection Technique .................................25 
Figure 14: Th?venin Equivalent of the Shunt Current Injection  
Technique at the Load ......................................................................................26 
Figure 15: Frequency Response of the 1
st
 Order Notch Filter ...........................................28 
Figure 16: Series Voltage Injection Technique .................................................................30 
Figure 17: Physical Connection of the Toroidal Coupler ..................................................31 
Figure 18: Th?venin Equivalent of the Series Voltage Injection  
Technique at the Substation .............................................................................32 
 viii
Figure 19: Shunt Voltage Injection Technique ..................................................................34 
Figure 20: Line Diagram of the Shunt Voltage Injection Technique ................................35 
Figure 21: Th?venin Equivalent of the Shunt Voltage Injection Technique  
at the Substation ...............................................................................................36 
Figure 22: Switched Impedance Current Signal ? on 1 cycle and off 9 cycles .................40 
Figure 23: FFT of the Switched Impedance Current Signal ?  
on 1 cycle and off  9 cycles ..............................................................................40 
Figure 24: Switched Impedance Current Signal ? on 5 cycles and off 5 cycles ................42 
Figure 25: FFT of the Switched Impedance Current Signal ?  
on 5 cycles and off 5 cycles .............................................................................42 
 
 
  
 ix
 
List of Abbreviations 
 
 
CT Current Transformer 
DAQ Data Acquisition 
EMC Electromagnetic Compatibility 
FFT Fast Fourier Transform 
FSK Frequency Shift Keying 
HV High Voltage 
LSB Least Significant Bit 
LV Low Voltage 
MV Medium Voltage 
PLC Power Line Communication    
1 
 
 
Chapter 1: Introduction 
 
Weather is unavoidable and often a very destructive force of nature.  High winds 
can knock trees over onto the ground.  The weight of ice and snow on tree limbs can cause 
them to snap.  These are a couple of the many weather conditions that can be detrimental 
to the distribution power grid.  It is the power company?s responsibility to keep the power 
line right of way clear of possible hazards.  However, storms do occur which sometimes 
result in down power lines and poles.  When this happens it is the job of the power 
company to restore the lost power to their customers as quickly as possible.  As loads are 
reconnected there exists a chance that a particular load could be reconnected to a different 
phase from which it was originally connected.  This results in an unbalanced power 
system (assuming the power system was previously balanced).  Knowledge of how loads 
are connected to the power grid is one of the ways the power company maintains a 
balanced power system.  Currently this knowledge is obtained by hiring people to ?walk 
the power lines? and make a table of the phase to which each load is connected.  This 
procedure is long, tedious, and most of all expensive, thus it is advantageous to implement 
a method that automates the process of load phase identification. 
There are several companies that produce devices that can communicate with each 
other and ultimately send their information to some master device. The primary 
communication protocol implemented by these devices is wireless communication.  This 
can be accomplished by using the existing cell phone infrastructure.  In the event a device 
 2
is out of range of a cell tower it is able to transmit its data to any available device that is 
within range.  If the second device is within range of a cell tower, then it will transmit its 
data and the data from the first device.  However, if it is also out of range of a cell tower, 
then it will transmit its data and the data of the first device to another available device.  
This process is called a mesh network [1].   
The wireless mesh network devices are primarily implemented when the desired 
data is meter data (e.g. power consumption, load voltage).  Due to the wireless nature of 
the device it does not automatically keep track of the phase to which the load is connected.  
It is possible to program the phase information into the device and have it included in the 
transmitted information packet.  However, the load could be swapped to a different phase 
at which point the device would have to be reprogrammed with the new phase data.  This 
still requires an individual to find the phase information and is not an automated process. 
An alternative approach is to use the power lines as the communication medium. 
This is referred to as power line communication (PLC).  The PLC scheme will allow for 
communication from a master device to several slave devices via the existing power lines 
and vice versa.  The master device will exist at a central location (e.g. a substation) and the 
slave devices will exist at the loads (e.g. meter boxes).  The slave will be assigned a 
unique identifier which will be communicated to the master.  A log will be developed by 
the master device as it receives an identifier from each slave device.  This log will have 
three sections that are based on the phases of the power system.  The master device will 
store the received identifier in the section of the log that corresponds to the phase that was 
used by the slave device as the communication medium.  An example of how the master 
device and slave devices are connected is provided in Fig. 1.  Each individual ID number 
 3
in Fig. 1 represents a slave device.  An example log developed by the master device that 
utilizes the information in Fig. 1 is provided in Table 1.   
 
 
The PLC scheme is also self healing.  If a load is disconnected from one phase and 
reconnected to another, then the slave device will transmit its identifier on the new phase 
to which it is connected.  The master device will receive the identifier on the new phase 
and update its log accordingly.  There is no longer a need for an individual to manually log 
how loads are connected as the PLC scheme has automated the process.  Balancing the 
power system is still a huge task but obtaining information regarding how loads are 
connected to the power system will make up a much smaller percentage of the work at 
hand.  
Phase A B C 
ID # 
1000 2182 3218 
2196 1308 1513 
1025 3015 1067 
Figure 1: Load Phase Connection Diagram 
Table 1: Example Data Log Based on Figure 1 
 4
 
Chapter 2: Communications 
 
A PLC scheme has two main activities that need to be resolved: a 
communications protocol and a signal coupling and decoupling technique to the power 
grid.  There are two primary communication strategies to consider as a communications 
protocol.  The first strategy involves the slave device verifying that the communication 
channel is open and then transmitting its data to the master device.  The second strategy 
requires the master device to send a request signal to the slave device.  The slave device 
will begin transmitting its data upon receiving the request signal from the master device. 
It is advantageous to transmit data using a frequency between 6-130 Hz [2].  The 
reason why this is the case is because of the system line impedance.  A power line?s 
impedance is a function of frequency.  The impedance has resistive, inductive, and 
capacitive elements.  As the communication signal?s frequency increases the impedance 
changes.  The inductive reactance increases because it is a function of frequency.  The 
resistive element increases because of the skin effect.  Current density moves to the edges 
of the wire as the frequency increases.  This reduces the area that the current is flowing 
through, which results in an increased resistance [4].  Capacitive reactance decreases 
because it is a function of the inverse of frequency.  High frequency operation changes 
the impedance which is likely to result in attenuation of the communication signal.   
The attenuation of the high frequency communication signal can be illustrated by 
utilizing an equivalent circuit model for the power line.  The power line model that is 
 5
typically employed is the positive sequence line pi model.  Unfortunately, this is a low 
frequency model and will not accurately describe high frequency operation because of 
errors associated with lumped-parameter models.  This is not necessarily an issue because 
the present model is sufficient to illustrate problems associated with high frequency 
communication with the power line as the communication medium.   
It is possible to develop a transfer function for this model.  The transfer function 
will provide information on how the power line impedance affects signals of various 
frequencies.  The line pi model is provided in Fig. 2.   
The transfer function that describes the ratio of the output voltage versus the input 
voltage as a function of frequency is described by H(?). 
 
nullnullnullnull null
1
nullnullnullnull
null
null1nullnull
null
null
null
null
. (1)
If the type of power line conductor and spacing information is known then the line 
parameter elements for (1) can be found.  As an example, assume the power line is a 477 
MCM 24/7 ACSR conductor that is 5 miles long with 3 feet of equilateral spacing [9].  
The resulting pi circuit model elements are a resistance R of 1.064 ?, an inductance L of 
10.91 mH, and capacitance?s C
1
 and C
2
 of 34.05 nF.  A frequency response plot of (1) 
that uses the calculated component values for the chosen type of power line conductor is 
provided in Fig. 3.  It is difficult to see how the transfer function responds as the 
Figure 2: Positive Sequence Power Line PI Model
 6
frequency increases.  The y-axis limits of Fig. 3 are adjusted in Fig. 4 to aid in the 
analysis of the frequency response as the frequency increases.  The transfer function 
magnitude approaches zero, indicating attenuation, as the frequency increases.   
Figure 4: Frequency Response of the Power Line PI Model with Adjusted Y-Axis Limit 
Figure 3: Frequency Response of the Power Line PI Model 
 7
This response is expected because of the skin effect, the increasing inductive reactance, 
and the decreasing capacitive reactance.  Notice the frequency range between 1-2 kHz in 
Fig. 4.  This model of the power system is amplifying the signal for this frequency range.  
This suggests that the communication signal?s frequency should be within this range as 
there is no attenuation.  However, the chosen power line model ignores several factors 
that exist in the real power system which prevent the communication frequency from 
being in the 1-2 kHz range.  An example is that a distributed line model and large power 
factor correction capacitor banks located at industrial facilities or on distribution feeders 
will introduce additional resonant points that would need to be addressed.  These power 
factor correction banks can introduce resonant points in the 180-200 Hz range and will 
certainly have resonant points by 300-420 Hz.  It is almost unheard of for there to be 
resonant points on the power line in the 6-100 Hz range.  A communication frequency in 
the range of 6-130 Hz is desirable because it avoids the issues that are associated with 
resonant points in the power system.  It is important to note again that the line pi model 
discussed here is not a good high frequency model.  Therefore, a low frequency 
communication signal is desired to keep attenuation of the communication signal at a 
minimum while avoiding network resonance conditions. 
 
  
 8
2.1: Slave Device Wait and then Transmit 
The first communication strategy involves the slave device waiting for the 
transmit line to become open and then transmitting the unique identifier to the master 
device.  The data that will be transmitted is the slave?s unique identifier, which is a binary 
number.  Line and transformer impedances make transmitting digital ones and zeros, 
based on 5 V logic, impossible.  It would be difficult for the master device to know when 
a zero is being transmitted and almost impossible to recognize a one.  The one would be 
transmitted as a 5 V pulse.  The pulse has infinite spectral content which would be 
filtered by the line and transformer impedances [5].  A suggested alternative would be to 
use sinusoids of a desired frequency to represent either a one or a zero.  The slave device 
would transmit a carrier frequency that would let the master know that it is about to 
transmit data.  After transmitting the carrier frequency, it would then shift the carrier 
frequency to a predetermined frequency that corresponds to a one or zero.  The slave will 
transmit the carrier frequency and then the frequency that corresponds to each bit until 
completion.  Once it has finished transmitting the data, it will resend the carrier frequency 
to let the master know it is complete.  The slave will transmit the carrier frequency and 
the frequency that represents each bit of data for a predetermined length of time, ?.  This 
process of using different frequencies for communication is called frequency shift keying 
(FSK) [3].  The master device would receive the frequency shifted sinusoid and calculate 
the frequency components of the signal to recover the transmitted data.  The other slave 
devices would need to check the channel for communication frequencies before 
transmitting.  If none are detected then transmission may begin.  This will ensure that the 
channel is open and will avoid communication errors due to multiple slave devices trying 
 9
to communicate at the same time.  An advantage of FSK is that it potentially allows for 
additional data to be transmitted with the unique identifier.  There is not a limit to the 
number of bits that can be transmitted; however, transmission of additional bits will 
increase the time that a particular slave device is communicating.  This time increase 
could be significant because the devices are communicating at low frequencies and there 
could be a larger number of devices all needing to communicate. 
An implementation example for this communication scheme is in order.  Assume 
that the desired data signal for transmission is binary 101.  A plot of the data signal is 
illustrated in Fig. 5.  The carrier frequency is chosen to be 6 Hz, while 3 Hz and 12 Hz 
are for a binary 0 and 1 respectively.  The slave device will first transmit the carrier 
signal for time ?, where ? is one second.  It is now time to begin data transmission.  A 
12Hz sinusoid is transmitted followed by a 3 Hz sinusoid and then finally a 12 Hz 
sinusoid is transmitted.  The 6 Hz carrier signal is retransmitted to indicate to the master 
device that data transmission is complete.  A plot of the transmitted signal is illustrated in 
Fig. 6.    
Figure 5: Data Signal
 10
 
 
2.2: Master Asks and Slave Responds 
A second communication strategy requires the master device to send out an 
identifier on all three phases.  The slave device will receive its identifier and respond.  It 
will respond with a sinusoid at some frequency (e.g. a tone) for a predetermined length of 
time.  The master will receive the tone and record the phase on which the tone was 
received.  This method avoids the potential problem of multiple slave devices trying to 
communicate at the same time because the master device initiates the communication.  
Unfortunately, this method severely limits the amount of information that can be 
communicated between the master and slave devices. 
 
  
Figure 6: Example of FSK
 11
 
Chapter 3: Coupling Introduction 
3.1: Signal Coupling 
Signal coupling involves connecting the communication hardware input and output 
to the power system.  There are two different situations that exist that have different 
constraints.  The first is that the slave device must be small enough to fit into the standard 
meter socket equipment.  This is problematic because as the slave device gets smaller the 
strength of the output signal may decrease.  This results in the device being unable to 
communicate as distance increases because the line impedance increases with line length.  
The second is that the substation equipment may have electromagnetic compatibility 
(EMC) issues.  It is possible that the injected signal levels could be large enough to 
compromise EMC for other equipment.  The working assumption adopted is that the meter 
socket size is a limiting constraint whereas the substation EMC issues can be managed.  
Therefore, the primary focus is to design a slave device that will fit into the standard meter 
socket equipment. 
 
3.2: Data Signal Strength for Successful Recovery 
It is important to know how big the data signal, from the slave device, needs to be 
for successful recovery by the master device.  A typical data acquisition (DAQ) card is 16 
bits.  If the card is set to a 10 V range, null5 V, then the value of the least significant bit 
(LSB) is about 150 ?V, from (2). 
 12
 
null
nullnullnull
null
10null
2
nullnull
null 152.6?V?
(2)
The smallest measurable data signal is taken to be 300 ?V.		This	allows	for	variations	of	
about	2*LSB	in	the	data	acquisition	hardware.  
The typical current supplied by a substation is about 200 A.  If 10 A is coupled 
onto the load conductor, at the standard meter, then there would be about 300 mA at the 
substation assuming a standard 7200:240 distribution transformer is used, from (3).   
 
null
null
null10null ?
240
7200
null 333.33nullnull?
(3)
If a standard 5000:5 winding CT is used for decoupling at the substation then the 
output signal will be 300 ?A from the communication signal and 200 mA from the power 
signal, from (4) and (5).   
 
null
null,nullnullnullnull
null 300nullnull ?
5
5000
null 300?null?
(4)
 
null
null,nullnullnullnullnull
null 200null ?
5
5000
null 200nullnull?
(5)
 
Assume that there is a 1 V per 1 A transducer connected to the output of the CT.  If this is 
the case then the master device must be capable of measuring a 300 ?V communication 
signal in the presence of a 200 mV power signal.   
As an example, the measured output of the transducer, with the 200 mV, 60 Hz 
signal and a 300?V, 130 Hz signal, could be filtered by a highpass chebyshev filter with a 
cutoff frequency of 60 Hz.  A fast fourier transform (FFT) would then need to be 
performed on the output of the filter.  The output of the FFT is illustrated in Fig. 7.  The 
130 Hz signal can be successfully recovered by the FFT.  Therefore, the coupled signal, at 
 13
the load, must be at least 10 A for successful recovery by the master device.  This 10 A 
level will be taken as a design requirement in this work. 
 
3.3: Assumptions 
The coupler design must be compatible with any service that meets flicker and 
voltage drop criteria.  The majority of available standard meters are used in residential 
applications.  Houses typically have a maximum current of around 100 A at 240 V.  The 
maximum voltage drop is 5%, usually based on flicker criteria.  If a motor starts and pulls 
100 A at 240 V then the Th?venin equivalent impedance magnitude seen by the load is 
0.12 ? for a 5% voltage drop from (6). 
 
null
nullnull
null
0.05? 240null
100null
null 0.12??
(6)
Figure 7: 130 Hz Signal Recovered by FFT
 14
The calculated impedance value includes the impedances of the line, distribution 
transformer, and substation equivalent impedance.  Also, a typical home has a minimum 
impedance magnitude of 2.4 ? when it is operating under maximum load from (7). 
 
null
nullnullnullnull
null
240null
100null
null2.4??
(7)
These assumptions are important because they set operation requirements for the master 
and slave devices. 
  
 15
 
Chapter 4: Coupling Techniques 
 
It is possible to couple either a voltage or a current onto the load conductor.  
There are four conventional techniques and one unconventional technique that were 
explored in coupling a communication signal onto the power line.  These techniques are 
series and shunt current injection, series and shunt voltage injection, and a switched 
impedance method.  Each of these techniques has their own advantages and 
disadvantages in coupling a communication signal onto the power line.  It is necessary 
for these coupling techniques to allow communication to occur at any time during the day 
and under any load condition. 
 
4.1: Series Current Injection Technique 
The series current injection technique will be discussed first.  This technique 
involves placing a toroid around the load conductor as illustrated in Fig. 8.  There is some 
number of windings placed on the toroid.   
Figure 8: Series Current Injection Technique
 16
The load conductor and the windings on the toroid act as a current transformer (CT) with 
a ratio of N:1 with N being the number of windings on the toroid.  A current is driven 
through the communication side of the CT which is then magnetically coupled onto the 
load conductor.  The coupled current is a series injected current.  One of the first 
concerns is coupling of the power signal back to the electronic side of the toroid.  If there 
is too much power signal transferred across the toroid, it will damage the communication 
hardware.  The current of the power signal that is reflected will be the ratio of the 
windings on the CT times the current through the CT on the load conductor.  If the load 
current is 100 A with N equal to 50 turns, then the communication hardware must be able 
to handle 2 A at the power frequency, from (8). 
 
null
nullnullnullnull
null 100null ?
1
50
null2null?
(8)
It is possible to decrease the current that the communication hardware has to handle by 
increasing N.  Some type of power frequency bypass filter, not shown in Fig. 8, would 
probably be used to manage the current. 
Core saturation is an issue with this approach at low frequency operation.  The 
magnetic flux, ?, is found by integrating the magnetic flux density B that is passing 
through a surface [4].  
 ??nullnull?nullnull? (9)
Equation (9) reduces to 
 ??nullnull? (10)
under idealized operating conditions where the magnetic flux density, B, evenly 
intersects the cross sectional area, A, of the toroid.  The magnetic flux is also equal to 
 17
 
?null
?
null
 
(11)
where ? is the flux linkage.  Faraday?s Law states that the time rate of change of the flux 
linkage results in an induced voltage, v, in the coil [6]. 
 
vnull
nullnull
nullnull
 
(12)
If the induced voltage is a sinusoid with amplitude, V
pk
, and radian frequency, ?, then 
solving (12) for ? gives 
 
?null
null
nullnull
null
sinnullnullnullnull. 
(13)
Solving for B using (10) and (11) results in 
 
Bnull
?
nullnull
. 
(14)
By using (13) and (14) the magnetic flux density for the toroid can be represented as  
 
Bnull
null
nullnull
nullnullnull
sinnullnullnullnull 
(15)
with units V?s/m
2
 or tesla, T.  A tesla is also equivalent to 10 kilo-gauss, kG. 
It can be seen from (15) that B is a function of the input voltage, operating 
frequency, number of windings, and core size.  Notice that the operating frequency is in 
the denominator of (15).  The amount of magnetic flux density increases as the operating 
frequency decreases; this drives the core into saturation based on Fig. 9.  The operating 
frequency will be limited to the range of 6-130 Hz because of the low pass filter nature of 
the power system.  Another pre-determined variable in (15) is the cross sectional area, A, 
which is controlled by the size of the core that is able to fit into the standard meter.  The 
input voltage, V
pk
, and the core windings, N, are the variables that can be manipulated to 
keep the core from saturating.  The magnetic flux density is related to the magnetic field 
 18
intensity, H, by the core relative permeability, ?
r
, and the permeability of free space, ?
0
, 
and is provided in (16) [4]. 
 Bnull?
null
?
null
H (16)
A common type of core material is a W-type material, which has a relative 
permeability of about 10,000.  Saturation curves for core materials are typically plotted as 
a B versus H curve or B-H curve.  The B-H curve for the W-type material is illustrated in 
Fig. 8 [7].  It is easy to see that a small change in B can quickly push the core from the 
linear region of operation into the saturation region. 
 
Electromagnetic saturation occurs when the magnetic core material is not able to 
contain all of the magnetic flux that is produced by a current that is flowing through 
windings that are wrapped around the magnetic material.  An alternate form of the 
Figure 9: B-H Curve of a W-type Material
 19
magnetic field intensity, H, for a toroid is provided in (17).  The variable l is the mean 
length path of the toroid which is roughly equal to its average circumference [4]. 
 
Hnull
nullnull
null
 
(17)
Assuming there is 1 turn placed onto a toroid with a mean length path of 10? cm with a 
10 A current flowing through the turn, then H is approximately  
 
Hnull
1?10null
10nullnullnull
null 31.8
null
null
null0.4nullnull. 
(18)
The result from (18) can be compared to Fig. 9 which is the B-H curve for a W-type 
material.  Notice that the plot is beginning to saturate at 0.4 Oe for a radiant temperature 
of 25 ?C and is saturated if the radiant temperature is 100 ?C.  It is important to note that 
the 10 A current that was chosen is necessary for successful recovery by the master 
device and the mean length path is such that the core would be almost too big to fit into 
the approximately 5.5 inch diameter standard meter socket.  If the mean length path is 
decreased then this will cause H to increase which will increase the core saturation.  Also, 
if there are more turns placed onto the core then the core saturation will increase.  
Unfortunately core saturation is a problem that is primarily due to the large current 
necessary for successful recovery by the master device and the small core material that is 
necessary to fit into the standard meter. 
Core saturation is a big issue with the series current injection technique; however, 
the primary issue is that performance cannot be guaranteed when the load is low to off 
(e.g. a house at night).  It is appropriate to ignore the core saturation issue for now 
because it is not the primary concern with this technique.  A line diagram of a typical 
 20
load is presented in Fig. 10 with a series connected current transformer.  The distribution 
line and the load conductor are considered to be ideal for this model.   
The Th?venin equivalent voltage and impedance (V
null
nullnull,nullnullnull
,Z
null
nullnull,nullnullnull
nullR
nullnull,nullnullnull
null
j?L
nullnull,nullnullnull
) are the open circuit voltage and open circuit impedance as seen at the 
distribution side of the substation towards the source.  Resistances, R
1
 and R
2
, and 
reactances, ?L
1
 and ?L
2
, are the non-ideal circuit elements of the distribution 
transformer.  The core elements for the distribution transformer are assumed to be 
infinite.  V
null
nullnullnullnullnullnull
 is the voltage supplied by the communication hardware to drive the 
necessary current into the shunt coupler for a 10 A output current.  V
null
nullnullnullnull
 is the 
communication voltage that exists when the required 10 A communication current is 
output by the CT.  The load is represented by the resistance R
L
 and reactance ?L
L
.  
Notice that the CT is modeled as an ideal transformer for this discussion.  This is 
Figure 10: Line Diagram of the Shunt Current Injection Technique 
 21
obviously not realistic, however the non-ideal factors (e.g. saturation) are ultimately not a 
limiting factor and are ignored. 
Series current injection involves magnetically coupling a current onto the load 
conductor.  Because this is the case, it is appropriate to simplify Fig. 10 and only look at 
the load impedance, the ideal coupler and the Th?venin equivalent impedance at the 
standard meter where the device is connected.  The simplified circuit is illustrated in 
Fig.11.  The coupled current from the CT will flow through the load impedance and 
through the Th?venin equivalent impedance.  Because the communication frequency is 
not 60 Hz it will not be affected by the source, thus the source is modeled as a short 
circuit.  Recall that it is desired that the load conductor current be at least 10 A for 
successful recovery by the master device at the substation. 
Figure 11: Th?venin Equivalent Line Diagram from Substation to Load 
 22
In order for 10 A to flow on the load conductor the voltage V
null
nullnullnullnull
 must be a 
factor of 10 greater than the load and Th?venin equivalent impedance from (19). 
 
nullnull
?
nullnullnullnullnullnullnull
nullnullnull
null
null
nullnullnullnull
null
?
nullnullnull
nullnullnull
null
null
nullnullnullnull
null
nullnull
nullnullnullnull
nullnull
nullnull
null
nullnullnullnull
null
nullnull10null (19)
The Th?venin impedance can be considered constant; however the load impedance will 
fluctuate.  If the load is operating at maximum power and then the power is decreased by 
50% then the corresponding load impedance will be increased by 50%.  This will require 
the communication voltage, V
null
nullnullnullnull
, to also increase by 50% so that the required 10 A is 
still coupled onto the load conductor.  This becomes problematic because V
null
nullnullnullnull
 has to 
increase by the same amount that the load impedance increases.  For small impedance 
variations this is not a very big problem, however when the load power significantly 
decreases or the load is turned off, then the impedance becomes very large.  It is very 
important to be able to control the magnitude of V
null
nullnullnullnull
 in the presence of the varying 
load impedance so that there is always 10 A coupled to the load conductor during 
communication.  Unfortunately it becomes impossible to drive 10 A for a no load 
condition because the load is now an open circuit. 
In order to couple the required communication current onto the load conductor it 
is necessary to be able to control the communication voltage, V
null
nullnullnullnull
.  The voltage V
null
nullnullnullnull
 
is related to the voltage at the communication hardware, V
null
nullnullnullnullnullnull
, by (20) because it is 
assumed that issues such as saturation are neglected.   
 
V
null
nullnullnullnull
null
1
N
V
null
nullnullnullnullnullnull
 
(20)
The communication hardware will be able to directly control the magnitude of V
null
nullnullnullnull
 
because of the linear relationship between V
null
nullnullnullnull
 and V
null
nullnullnullnullnullnull
 that is presented in (20).  
 23
Unfortunately, the maximum magnitude of V
null
nullnullnullnullnullnull
 is limited by the low voltage nature of 
the communication hardware.  As the load impedance increases a threshold will be 
crossed where V
null
nullnullnullnull
 cannot be increased enough to supply the necessary 10 A current 
because V
null
nullnullnullnullnullnull
 has reached its maximum magnitude.   
An application example to calculate V
null
nullnullnullnullnullnull
 is in order.  Assume that the Th?venin 
impedance is a short circuit and the magnitude of the load impedance is 2.4 ?.  Also 
assume that there are 5 turns on the communication side of the coupling CT.  By utilizing 
(19) and (20) it is possible to calculate that V
null
nullnullnullnullnullnull
 has to be 120 V in order to couple the 
required 10 A onto the load conductor.  This calculated voltage will increase as the load 
impedance increases.  It could be difficult to produce 120 V because of the low voltage 
nature of the communication hardware but it is possible.  Now consider a situation where 
the load impedance is 24 ? instead of 2.4 ?.		This	would	require	the	low	voltage	
communication	 hardware	 to produce 1.2 kV which would be problematic if not 
impossible. 
The series current injection coupling method is therefore an unrealizable PLC 
coupling method.  This is due to both the saturating nature of the small core and the 
required potentially large voltage at the low voltage communication hardware.  While 
core saturation is an issue that will have to be considered it is not the primary problem 
with this methodology.  In fact, saturation issues were ignored while discussing the 
necessary voltage magnitude to couple the 10 A communication signal.  The series 
current injection technique is an unrealistic coupling methodology because of the large 
voltages that must be supplied by the communication hardware in order to always be able 
to couple the required communication current onto the load conductor.  
 24
4.2: Shunt Current Injection Technique 
The second technique to be discussed is the shunt current injection method.  This 
technique involves connecting a conventional instrument CT in shunt with the load.  The 
communication hardware is connected to the secondary side of the CT.  A 60 Hz 
blocking filter will be connected, in series, with the shunt connection.  The blocking filter 
will keep the shunt connection from shorting out the load at the power frequency.  A 
diagram of how the components are connected is illustrated in Fig. 12.  Unfortunately, 
the core saturation issues that were discussed for the series current injection technique 
will still apply to this method.  However, this method does have an advantage over the 
series current injection method, in that the load impedance does not severely impact the 
coupled current. 
It is appropriate to analyze the circuit provided in Fig. 12 and compare the results 
to the series current injection method.  The equivalent circuit line diagram for Fig. 12 is 
illustrated in Fig. 13.  The distribution line and the load conductor are considered to be 
ideal for this model.  The Th?venin equivalent voltage and impedance (V
null
nullnull,nullnullnull
,
Z
null
nullnull,nullnullnull
nullR
nullnull,nullnullnull
nullj?L
nullnull,nullnullnull
) are the open circuit voltage and open circuit impedance 
Figure 12: Shunt Current Injection Technique
 25
as seen at the distribution side of the substation towards the generator.  Resistances, R
1
 
and R
2
, and reactances, ?L
1
 and ?L
2
, are the non-ideal circuit elements of the 
distribution transformer.  The core elements for the distribution transformer are assumed 
to be infinite.  V
null
nullnullnullnullnullnull
 is the voltage supplied by the communication hardware to drive the 
necessary current into the shunt coupler for a 10 A output current.  V
null
nullnullnullnull
 is the 
communication voltage that exists when the required 10 A communication current is 
injected by the CT.  The load is represented by the resistance R
L
 and reactance ?L
L
.  
Notice that the CT is modeled as an ideal transformer for this discussion.  The non-ideal 
elements are not a limiting factor for this coupling scheme and are ignored. 
The circuit provided in Fig. 13 can be further reduced by making a Th?venin 
equivalent at the low voltage (LV) side of the transformer.  This removes the distribution 
transformer from the circuit and simplifies the analysis of the system modeled in Fig. 13.  
The new simplified model is illustrated in Fig. 14.  The Th?venin equivalent impedance 
in Fig. 14 is modeled by resistance R
TH,L
 and reactance ?L
TH,L
. and the Th?venin voltage 
is modeled as a short at the communication frequency. 
Figure 13: Line Diagram of the Shunt Current Injection Technique 
 26
The output current from the shunt coupler needs to be at least 10 A in order for 
there to be successful communication between the master and slave devices.  This current 
is calculated in (21).  An approximate Th?venin impedance magnitude was found in (6).  
The minimum load impedance would be 2.4 ? for a load operating at 240 V and 100 A.  
The calculation in (21) can be further reduced by ignoring the load impedance because it 
is at least an order of magnitude larger than the Th?venin impedance.  The new current 
equation is provided in (22). 
 
 
nullnull
?
nullnullnullnullnullnullnull
nullnullnull
null
null
nullnullnullnull
nullnull
nullnull,null
nullnullnullnull
nullnull,null
null||nullnull
null
nullnullnullnull
null
null
null (21)
 
nullnull
?
nullnullnullnullnullnullnull
nullnullnull
null
null
nullnullnullnull
null
nullnull,null
nullnullnullnull
nullnull,null
null (22)
 
This method does not have the high voltage implications at the communication 
hardware when compared to the series current injection technique.  Substituting 10 A and 
the results from (6) into (22) will result in the necessary voltage output from the shunt 
coupler for successful communication between the master and slave devices.  The 
Figure 14: Th?venin Equivalent of the Shunt Current Injection Technique at the Load 
 27
calculated voltage is 1.2 V for the 0.12 ? Th?venin impedance found in (6).  If there are 5 
turns on the shunt coupler then the voltage V
null
nullnullnullnullnullnull
 will be 6 V from (20).  This result is 
much more realistic than the 100 V (or more!) result from the series current injection 
technique.  The low voltages necessary at the communication hardware, for 
communication between the master and slave device, make this technique much more 
realistic to implement than the series current injection technique.  
The shunt current injection technique removes the problem involving a high 
voltage at the communication hardware that plagues the series current injection 
technique.  However, there is a new issue with the shunt current injection technique that 
needs to be addressed.  A series filter is necessary with the shunt coupler to prevent the 
60 Hz power signal from short circuiting.  This filter will also need to fit into the standard 
meter socket with the shunt coupler. 
A notch filter is desired to block the 60 Hz power signal from short circuiting 
when the shunt coupler is connected.  The filter will need to block a 240 V
RMS
 signal 
because that is the voltage present at the standard meter.  It is necessary for the filtered 
magnitude of the 60 Hz power signal to be no bigger than the voltage that is produced by 
the 10 A coupled current.  The actual filtered magnitude will probably need to be much 
smaller than the voltage produced by the 10 A coupled current.  This is to prevent the 
sum of the 60 Hz signal and the produced voltage from exceeding any current or voltage 
limits that are present at the communication hardware. 
 28
It is very likely that the components that are used to make the notch filter will be 
too large to fit into the standard meter socket.  A first order notch filter can be considered 
to see how big the passive components are that comprise the filter.  The magnitude of the 
frequency response, Z(?), for a first order notch filter is illustrated in Fig. 15.  The 1
st
 
order notch filter does an excellent job of filtering out the 60 Hz power signal and is very 
easy to implement.   
 
Implementing the filter requires an inductor and a capacitor to be in parallel with 
each other.  These parallel components are then connected in series with the 
communication hardware.  The size of the inductor is selected as 3.2 ?H and the size of 
the capacitor is calculated to be 2.2 F.  A 2.2 F capacitor with a voltage rating of 
240V
RMS
 is not available for purchase. This can be solved by connecting multiple 
Figure 15: Frequency Response of the 1
st
 Order Notch Filter 
 29
capacitors in parallel to obtain the required 2.2 F capacitance.  A typical 250V
DC
 
capacitor (which is only considered to give an idea of the physical size requirement) will 
have a capacitance range between 470 ?F to 38 mF.  It will take approximately 58 of the 
38 mF capacitors in parallel to create an equivalent capacitance of 2.2 F.  Each of these 
38 mF capacitors is 3.5 inches in diameter and 8.625 inches long [8].  It would be 
impossible to fit this number of capacitors with these dimensions into the standard meter 
socket so as to implement the notch filter for the shunt current injection technique.  Other 
combinations of L and C values would lead to similar size problems. 
The shunt current injection technique is unlikely to be implemented because of 
the size issue with the 60 Hz notch filter.  The components that are necessary to build the 
filter will be too large to fit into the standard meter socket; one of the design requirements 
would clearly be violated.  Even though the shunt current injection technique solves the 
problem involving high voltages at the communication hardware, it is an undesirable 
coupling strategy due to component size. 
  
 30
4.3: Series Voltage Injection Technique 
Series voltage injection is the third coupling technique covered.  This technique 
involves placing a voltage transformer in series with the load.  The communication 
hardware is connected to the secondary side, N2, of the transformer.  Some type of power 
frequency bypass filter is likely to be required but it is not depicted in Fig. 16 because it 
is not relevant to the discussions that follow.  A diagram of how the components are 
connected is illustrated in Fig. 16.   
The major drawback to this method is that the primary winding, N1, of the transformer 
has to carry the full load current.  In order for the primary side to carry the load current, it 
has to be wound with wire that is large enough to carry the current.  Winding the 
transformer with large wire will result in a large transformer that will not fit in the 
standard meter socket.  This coupling method is undesirable because the size of the 
coupler will be too big for the application. 
An alternate option is to utilize a toroid as the magnetic material of the voltage 
transformer.  This implementation is similar to the series current injection technique; 
however the emphasis is on coupling a voltage instead of a current.  The load conductor 
will be passed through the toroid which will make the primary winding, N
1
, equal to one.  
Figure 16: Series Voltage Injection Technique
 31
An example of the physical connection of the coupler is illustrated in Fig. 17.  The 
shaded area represents the secondary windings, N
2
, that are placed on the coupler.   
In order for there to be successful communication the voltage that is dropped 
across the Th?venin equivalent source impedance at the substation needs to be large 
enough to be recovered by the master device.  The Th?venin equivalent source 
impedance for the substation is unknown.  It is appropriate to assume that the impedance 
magnitude found in (6) is the Th?venin equivalent source impedance, Z
TH
?, at the 
substation.  The actual value of Z
TH
? will be less than the value provided in (6) because 
(6) includes the distribution line and distribution transformer impedances.  Note that the 
assumed equivalent source impedance is reflected onto the LV side of the distribution 
transformer and is denoted by the prime.  Reflecting Z
TH
? to the high voltage (HV) side 
of the distribution transformer will result in an impedance of 108 ? for Z
TH
.  Z
TH
 is the 
assumed magnitude of the Th?venin impedance at the substation towards the source. 
Figure 17: Physical Connection of the Toroidal Coupler 
 32
An equivalent circuit model is provided in Fig. 18.  This circuit model ignores the 
distribution transformer and distribution line impedances.  It also ignores impedances 
associated with the series voltage coupler and assumes that the transformer magnetic 
material is infinitely permeable.  The source voltage is modeled as a short because the 
communication frequency is not 60 Hz. 
The max voltage drop across Z
TH
 will occur when the load impedance, Z
L
, is at a 
minimum.  If Z
L
 is 2.4 ?, from (7), then V
nullnullnullnull
 can be calculated as a function of the 
input signal voltage V
nullnullnullnullnullnull
.  
 null
nullnullnullnull
null
null
nullnullnullnullnullnull
null
null
null
null
nullnull
null
nullnull
nullnull
null
null
7200
240
null null
null
nullnull
7200
240
null  (23)
Figure 18: Th?venin Equivalent of the Series Voltage Injection Technique at the Substation
 33
A typical data acquisition unit has 16 bits.  The medium voltage (MV) power signal is 
assumed to be 7200 V
RMS
 line to neutral.  Each bit will represent 311 mV for the 
?10.2kV or 20.4 kV input range from (24).  The smallest measurable data signal is taken 
to be 622 mV.  This allows for an error of 2*LSB in the DAQ unit?s hardware. 
 
null
nullnullnull
null
2?null?2 ? 7200null
2
nullnull
null 311 nullnull 
(24)
An application example is in order to see how the varying load impedance will 
affect the output communication signal.  Assume that the secondary side of the series 
coupler is wound with 50 turns and that the magnitude of V
nullnullnullnullnullnull
 is 5 V.  The output 
voltage V
nullnullnullnull
 is equal to 143 mV for a load impedance of 2.4 ?.  This voltage is below 
the smallest measurable signal and will be unrecoverable by the master device. 
The main problem with the toroid approach to the series voltage injection 
technique is the load impedance.  When the load is operating at max power the voltage 
that is at the substation is too small to be measured by the master device.  Note that it is 
assumed that there are not any magnetic saturation issues with the core material.  The 
coupled voltage will be dropped across the load impedance instead of the substation 
equivalent impedance.  This results in a data signal that cannot be detected by the master 
device unless the communication voltage is unacceptably high. 
There were two approaches that were discussed for implementing the series 
voltage injection technique.  The first involved placing a voltage transformer in series 
with the load.  This approach is problematic because the primary winding of the 
transformer will have to carry the full load current.  Wrapping wire that is large enough 
to carry the full load current will result in a large transformer.  The second approach was 
to use a toroid as the magnetic coupler and to place the load wire through the toroid.  This 
 34
approach removes the problem of a large transformer that is due to large windings; 
however its problem is the load impedance.  Even when the load is operating at max 
power (minimum impedance), its impedance prevents communication between the master 
and slave devices.  In summary, the series voltage injection technique is an undesirable 
technique because of either large transformer windings or load impedance. 
 
4.4: Shunt Voltage Injection Technique 
The shunt voltage injection is the last conventional coupling technique discussed.  
This technique involves placing a conventional transformer in parallel with the load.  The 
communication hardware will put a voltage signal onto the secondary side of the 
transformer which will then be coupled to the primary side.  Some type of power 
frequency blocking filter is likely to be required but is not shown because it is not 
relevant to the discussions that follow.  A diagram of how the components are connected 
is illustrated in Fig. 19.   
Trouble arises due to impedance issues with this technique.  The coupling 
transformer has a high impedance relative to the Th?venin equivalent impedance of the 
source at the substation.  An equivalent circuit of Fig. 19 is illustrated in Fig. 20.  The 
Figure 19: Shunt Voltage Injection Technique
 35
Th?venin equivalent voltage and impedance (V
null
nullnull,nullnullnull
,Z
null
nullnull,nullnullnull
nullR
nullnull,nullnullnull
nullj?L
nullnull,nullnullnull
) 
are the open circuit voltage and open circuit impedance as seen at the substation towards 
the source.  The distribution line impedance and transformer equivalent circuit impedance 
are assumed to be ideal.  The load impedance is represented by resistance R
L
 and 
reactance ?L
L
 and the load conductor impedance is assumed to be ideal.  The resistances, 
R
1
 and R
2
, and reactances, ?L
1
 and ?L
2
, are the equivalent circuit elements for the shunt 
coupling voltage transformer.  The equivalent shunt elements of the coupling transformer 
equivalent circuit are assumed to be very large and are disregarded. 
The issue arises due to the nature of the coupling transformer.  A low power 
signal transformer will have numerous windings on the primary side with a small gauge 
wire.  The large number of windings translates into a long wire that will have a large 
resistance which will make R
1
 a large value.  The resistance R
1
 is not the total resistance 
that will be present.  By reflecting R
2
 across the transformer to the primary side, it is 
possible to find the total resistance of the coupler which includes both R
1
 and R
2
. 
It is possible to calculate the magnitude of V
null
nullnullnullnullnullnull
 that is necessary for successful 
communication with the master device.  The Th?venin equivalent impedance at the 
Figure 20: Line Diagram of the Shunt Voltage Injection Technique 
 36
substation is unknown.  It is appropriate to make the same assumption for this discussion 
as was made in the series voltage injection technique.  The assumption is that the 
Th?venin equivalent impedance at the substation is the same as the impedance found in 
(6).  Note that the impedance in (6) is on the LV side of the distribution transformer.  It 
needs to be reflected to the HV side of the distribution transformer to apply to the 
equivalent circuit in Fig. 20.  This calculation is provided in (25). 
 
Z
nullnull,nullnullnull
null
7200
null
240
null
? 0.12? null 108? 
(25)
A small transformer with a primary to secondary winding ratio of 480 to 4 will have a 
resistance of about 4 k? for R
1
 and a resistance of about 0.5 ? for R
2
.  To simplify 
calculations the coupler reactances ?L
1
 and ?L
2
 and resistance R
2
 are ignored.  The new 
simplified model involves the Th?venin impedance at the substation, an ideal distribution 
transformer, the coupling transformer?s primary winding resistance, and the load 
impedance.  The Th?venin source voltage is a short because the communication 
frequency is not 60 Hz.  A circuit of the model is illustrated in Fig 21.    
Figure 21: Th?venin Equivalent of the Shunt Voltage Injection Technique at the Substation
 37
The communication voltage, V
nullnullnullnull
, can now be found as a function of the input voltage 
V
nullnullnullnullnullnull
.  The equation describing this relationship is provided in (26). 
 
V
nullnullnullnull
nullV
nullnullnullnullnullnull
null
null
null
null
null
nullnull
7200
240
nullnull
null
nullnull,nullnullnull
null
nullnull,nullnullnull
nullnull
null
null
7200
240
null null
null
null (26)
If the maximum voltage output by the communication hardware is 5 V then the 
communication voltage calculated in (26) will be 540 mV for a communication 
transformer turns ratio of 480 to 4.  This calculated voltage does not meet the 622 mV 
minimum voltage requirement for successful recovery by a 16 bit DAQ.   
There were several assumptions made to maximize the output voltage.  If all of 
the impedances that were omitted were accounted for in (26), then the communication 
voltage would be decreased further.  It is possible to overcome the attenuation by 
increasing the signal voltage level.  However, this is also problematic because higher 
voltages equates to larger components.  These larger components will not fit into the 
standard meter thus breaking the size requirement. 
One possible solution would be to increase the Th?venin equivalent impedance at 
the substation for the desired communication frequency.  There could be sufficient space 
at the substation to build a device that has a large impedance at the communication 
frequency.  If this were implemented then the impedance Z
TH,sub
 in (26) could be adjusted 
such that a desired communication voltage could be measured.  Unfortunately this device 
will be complex and expensive to build which will make implementation unlikely. 
The shunt voltage injection technique avoids the varying load impedance problem 
that plagues the series voltage and series current injection techniques.  However, the 
attenuation of the communication signal by the coupling transformer will render the data 
 38
signal unrecognizable by the master device.  This could be resolved by implementing a 
device that increases the Th?venin equivalent impedance at the substation for the desired 
communication frequency.  Implementation of the device is unlikely because of cost and 
complexity.  In summary, utilization of the shunt voltage injection technique is 
improbable as the communication signal will likely be too small to be successfully 
recovered by the master device.  Of all the magnetic coupling options, however, this one 
appears to have the greatest potential.  
 39
4.5: Switched Impedance Method 
The last technique covered is the unconventional switched impedance method.  
This method involves switching an impedance on/off at some frequency to produce a 
desired transmit signal.  In this scheme, the master device will transmit a slave?s 
identifier that the slave device will receive.  When the slave device receives the signal it 
will begin switching an impedance.  This switched impedance will draw current that will 
have spectral content at the desired transmit frequency.  The master will detect the 
content at the communication frequency and record the phase on which it was received.  
It is important to note that the transmitter encounters the same coupling challenges 
previously discussed, but it is assumed that solutions are possible because size is not an 
issue at the substation where the master transmitter is located.   
There are two examples of the switching technique presented for discussion.  The 
first example involves switching the impedance on for one cycle of the 60 Hz power 
signal and then off for 9 cycles.  The second example involves switching the impedance 
on for 5 cycles and off for 5 cycles.  Both of these examples will produce a current on the 
power line that will have spectral content at the desired example transmit frequency of 
6Hz. 
The first example involves a 60 Hz power signal being applied to an impedance 
for 1 cycle and then disconnected for 9 cycles to produce a current.  The resultant current 
is illustrated in Fig. 22.  An FFT of the current in Fig. 22 is illustrated in Fig. 23.  The 
FFT recovers a 6 Hz component and its magnitude is almost 1 A.  The 1 A component is 
an order of magnitude lower than the required 10 A for signal recoverability by the 
master device.  Also, the 50 A current peak in Fig. 22 is too big for the communication 
 40
hardware and could create undesirable short-term voltage fluctuations.  Implementing 
this particular switching method to produce a 6 Hz current is undesirable because of the 
large 60 Hz currents necessary to establish communication. 
  
Figure 23: FFT of the Switched Impedance Current Signal ? on 1 cycle and off 9 cycles 
Figure 22: Switched Impedance Current Signal ? on 1 cycle and off 9 cycles 
 41
In the second example the 60 Hz power signal is applied to an impedance for 5 
cycles and then disconnected for the next 5 cycles to produce a current.  The resultant 
current is illustrated in Fig. 24.  An FFT is performed on the current to verify the 
existence of the desired frequency component that is necessary for communication.  The 
result of the FFT is illustrated in Fig. 25.  There is a 6 Hz component recovered with a 
magnitude of approximately 1 A, however this signal is too small for successful 
communication to be established with the master device. 
Both of the discussed examples are successful in getting a 6 Hz current onto the 
power line.  The primary issue, with both examples, is the large 60 Hz current that is 
necessary to produce a 6 Hz component that the master device will be able to recover.  
Both examples involve a large 60 Hz current to produce a 6 Hz component with a 
magnitude of 1 A.  Unfortunately, a 10 A magnitude is necessary (assuming no special 
CTs at the substation), which would require the 60 Hz current peaks in Fig. 24 and Fig.25 
to be increased by a factor of 10.  These current peaks will be too large for the 
communication hardware and would almost certainly create other problems as well.  This 
technique is unlikely to be implemented because of the large 60 Hz current that is 
necessary to produce a 6 Hz component in the power line. 
 42
  
Figure 25: FFT of the Switched Impedance Current Signal ? on 5 cycles and off 5 cycles 
Figure 24: Switched Impedance Current Signal ? on 5 cycles and off 5 cycles 
 43
 
Chapter 5: Summary 
 
It is desirable to establish communication between a central location (e.g. 
substation) and the loads that are supplied by the source.  Successful communication 
involves two main processes.  The first process is to establish a communication protocol 
that the devices will be able to interpret.  The second process is a coupling scheme to 
connect the communication signal to the power line. 
There were two types of communication strategies discussed.  The first required 
the slave device to wait until the transmission channel is open before it begins 
transmission.  The second method involves the master device sending out the slave?s 
identification number on all three phases of the power system.  When the slave receives 
its identification number it responds to the master device. 
Several methodologies were presented to couple the data signal onto the power 
line to establish communication between the master and slave devices.  It is apparent 
from this research that it is difficult to couple a signal onto the load conductor.  Low 
frequency signal coupling is not straightforward due to the magnetic saturation issues 
arising from using cores that will fit into the standard meter socket.  The series current 
injection technique is undesirable due to the varying load impedance.  Shunt current 
injection is not going to work well because of the large size of the required power 
frequency blocking filter.  The series voltage injection method is limited by the primary 
coupling transformer winding having to carry the full load current or the large load 
 44
impedance.  Shunt voltage injection will not work well because of the signal attenuation 
by the coupling transformer.  The most promising technique is the switched impedance 
method.  However, it is unlikely to be implemented because of the large currents at the 
power frequency that are required to get the necessary low frequency current at the 
transmit frequency.  
  
 45
 
Chapter 6: Recommendations and Future Work 
 
The idea of establishing communication between a master and slave device is still 
of interest.  Using magnetics to couple a signal onto the load conductor should be 
avoided.  A directly coupled signal generator operating at the desired data transmission 
frequency should be implemented.  This can be accomplished by using an ?active filter? 
in the standard meter socket.  Once there is active communication at the load side, the 
?master ask then slave respond? methodology appears to be best suited.  Conventional 
coupling techniques should be used with the master device for communication as there is 
enough space available to avoid a directly coupled scheme.  All of the previous issues at 
the load side are also issues at the source side in relation to coupling.  However, the 
issues at the source side will be manageable because there is adequate space available for 
the large components that may be necessary for coupling and decoupling signals onto or 
from the power line.  
  
 46
 
References 
 
[1] Sensus, ?Licensed versus Unlicensed Spectrum for Utility Communications ? 
WP-100,? 2010. 
[2] Paul C. Hunt and Lynn R. Hunt, Using Ultra Narrow Bandwidth to Overcome 
Traditional Problems with Distribution Line Carrier, Proc. Rural Electric Power 
Conference, pp D3/1 - D3/8, April 1995. 
[3] B. P. Lathi, Modern Digital and Analog Communications Systems, 3rd ed., 
Oxford University Press, 1998. 
[4] Stuart M. Wentworth, Applied Electromagnetics: Early Transmission Lines 
Approach, Wiley 2007. 
[5] Sanjit K. Mitra, Digital Signal Processing: A Computer-Based Approach, 3rd ed., 
McGraw-Hill 2006. 
[6] J. David Irwin and R. Mark Nelms, Basic Engineering Circuit Analysis, 8th ed., 
Wiley 2005. 
[7] Magnetics, ?Materials ? Section 3,? 2011. 
[8] CDM Cornell Dublier, ?Type DCMC 85 ?C High Capacitance, Screw Terminal, 
Aluminum,? 2011. 
[9] Alcan Cable, ?Bare Overhead Transmission and Distribution Conductors ? 
UT0003,? January 2008.