Automotive Power Line Communication: A New Wiring Topology for Powertrain Sensor Network

(1)

Automotive Power Line Communication:

A New Wiring Topology for Powertrain

Sensor Network

A pre-study on the technical feasibility of implementing power line communication for Volvo powertrain sensor network

TEEMU TUORINIEMI

Masters’ Degree Project

Stockholm, Sweden 2013

(2)

(3)

“Electricity. The power that causes all natural phenomena not known to be caused by something else”

(4)

by Teemu Tuoriniemi

In motor vehicles the power and information are distributed by their own separate wires and the amount of wires over the past decades has increased tremendously due to the ever increasing electronics onboard. Solely the average weight of the wires has increased from 4 to 91 kg over the last 50 years and it is therefore clear that a remedy to the ever increasing wiring is needed. A possible solution to this could be to introduce power line communication (PLC) onboard a vehicle. The PLC aims at overlaying the information on top of the already existing power feed cables and thus eliminating the need of dedicated wires for communication purposes. The PLC solution would thus simplify the wiring network to a bare minimum, since no additional wires besides the power cables are needed.

During this thesis work the PLC technology was studied as a possible cost and quality reform of the powertrain sensor network, where it could be used to both reduce the cost of wires and increase the system reliability.

A theoretical background study was first performed to investigate the limitations and possibilities of the PLC implemented in a vehicle, and the PLC technology was also tested upon the DC-lines of an e6 Volvo FMX truck, between the post catalytic NOx

sensor and the ACM with two CAN protocol based PLC modems. The throughput, noise and the scalar voltage gain was measured in this link with the ignition key at different positions to test different modes of operations. It was shown that this particular link had a clear low pass characteristics with severe voltage attenuation without any significant difference caused by the different modes of operation. The severe characteristics of the link did result into a fault confinement mode of the CAN based PLC modems where their communication capabilities were inhibited, which shows that at least with this particular link, reliable communication was not possible with the equipment at hand. Short EMC measurements were also carried out regarding the radiated emissions. The results showed that the PLC technology implemented in the existing wiring architecture could be a possible cause of EMC problems if no counter actions are been taken into account.

(5)

Acknowledgements

I would like to thank Volvo for giving me the opportunity to do my Master thesis with them. The time spent here has truly been educational and the things I have learned will certainly be put into use in my future work.

The persons I would like to thank are first and foremost the people that has supervised this thesis: Patrik Andreasson, Carl-Evert Olsson, PhD and Lisa Nystr¨om, without whom this thesis would not been possible. I would like to thank Antonio La Sala, Jim Fischer and Sven Leicher-Ehl for all the help with cables, wires and with the economical aspects. Magnus Einarsson for all the help with the ACM and break out connections. Carl-Eric Bengtsson for the EMC discussions and for the assistance with the measure-ment devices, and Ulf Herbertsson for all the assistance with the radiation measuremeasure-ments in the anechoic chamber. I would also like to thank all the people working in BF93313 group from and during the time of this thesis without whom the work would not have been as interesting and worthwhile as it has been.

Finally I would like to thank Anna for her patience and support during the time of this work.

(6)

(7)

List of Figures

2.1 Communication over AC . . . 9

2.2 Communication over DC . . . 9

2.3 PLC blockdiagram . . . 10

2.4 engine wire harness . . . 12

2.5 engine wire harness . . . 12

2.6 Illustration of loads connected to the DC-lines that could cause reflections due to improper impedance matching. . . 14

2.7 Transmission line with the input impedance, Zg and load impedance ZL . 15 2.8 Input impedance . . . 16

2.9 Input impedance matched . . . 17

2.10 Power ratio . . . 19

2.11 balanced signaling . . . 21

2.12 Unbalanced signaling. . . 21

3.1 SIG60 . . . 29

3.2 DCAN250 . . . 29

3.3 Signals along the PLC link using DCAN250. Yellow signals are CAN signals and blue signals are modulated PLC signals . . . 30

3.4 PLC signal along the SIG60 . . . 31

3.5 Spectral content of the SIG60 signal . . . 31

4.1 The link . . . 35

4.2 Break out connection for NOx . . . 36

4.3 Break out box for the ACM . . . 36

4.4 Setup . . . 37

4.5 Frequency sweep . . . 38

4.6 Reciprocity measurements . . . 39

4.7 s12 during different conditions. . . 40

4.8 s12 measurements made with DCAN250 . . . 41

4.9 Noise level . . . 42

4.10 CANanalyzer screenshot . . . 43

4.11 CAN communication results from CAN1 to CAN2 . . . 46

4.12 CAN communication results from CAN2 to CAN1 . . . 46

5.1 CAN-to-optical transducer and DCAN250 . . . 50

5.2 The test setup for EMC measurements . . . 50

5.3 Average EM-field . . . 51

5.4 Peak EM-field . . . 51

(12)

A.1 Transmission line . . . 67

(13)

List of Tables

4.1 PLC test results engine idling CAN2 to CAN1 . . . 44

4.2 PLC test results truck in pre-running state, CAN2 to CAN1. . . 44

4.3 PLC test results engine idling, CAN1 to CAN2 . . . 44

4.4 PLC test results with the truck being at the pre-running state, CAN1 to CAN2 . . . 45

4.5 PLC test results with the engine idling and with a separate return con-ductor, CAN2 to CAN1 . . . 45

4.6 PLC test results while truck is at pre-running state and with a separate return conductor, CAN2 to CAN1 . . . 45

4.7 PLC test results while the engine is idling and with a separate return conductor, CAN1 to CAN2 . . . 46

4.8 PLC test results while truck is at pre-running state and with a separate return conductor, CAN1 to CAN2 . . . 46

6.1 The wires that can be removed from EATS-harness if PLC is introduced. 56

A.1 Distributed parameter equations . . . 70

A.2 Distributed parameter values . . . 71

(14)

(15)

Abbreviations

CAN Control Area Network LIN Local Interconnect Network EMC ElectroMagnetic Compatibility EMI ElectroMagnetic Interference ECU Electronic Control Unit PLC Power Line Communication TEM Transverse ElectroMagnetic EMS Engine Management System ACM After treatment Control Module FFT Fast Fourier Transform

LISN Line Impedance Stabilization Network EATS Engine After Treatment System

(16)

(17)

Physical Constants

Speed of Light c = 2.997 924 58 × 108 ms−s Free Space Permittivity 0 = 8.8541878176 × 10−12Fm−1

Free Space Permeability µ0 = 4π × 10−7 Hm−1

(18)

(19)

Symbols

f Frequency s−1

B Magnetic flux density T E Electric field strength Vm−1 J Current density Am−2 v Speed ms−1 R Resistance Ω C Capacitance F L Inductance H G Conductance Ω−1 I Current A V Voltage V Z Impedance Ω l Length m t Time s P Power W

ω Angular frequency rad·s−1

(20)

(21)

Part I

Introduction

(22)

(23)

Chapter 1

Introduction

1.1 Background

In motor vehicles the average amount of wires has increased from 4 to 91 kg in the last 50 years [19] because of the ever increasing electronics onboard. Today in a modern Volvo truck the wiring harness has become the second heaviest component onboard after the engine and the amount of wires has increased the thickness of the wiring harness to cause problems with the installation. The problems are not just limited to weight and size but also the electromagnetic compatibility (EMC) issues has become more and more important [6].

The incentives become clear that the wiring needs to be reduced, and this does not only concern the automotive industry but also aviation and space industries as well [5] [10]. With every kilogram, meter or the amount of connection points that could be reduced there could be a significant reduction of the cost due to the price of wire and installation time. For every avoided kilogram in the vehicle there will be an increase in performance and the overall reliability, since less weight will increase the fuel economy and with less wire connections there will be fewer faults that could occur.

The ultimate goal would be to avoid wiring altogether, but the wireless technology is not always the best solution since the difficulties of power supply, that is if a sensor or another end device cannot harvest their own energy, power lines would be routed to these nodes in either way. Thus a suggested half-way solution to this is the introduction

(24)

of power line communication (PLC). PLC aims at superimposing the information on top of the already existing power feed cables and since today the power and information is routed through their own separate cables, the PLC solution offers the possibility of removing the information cables altogether.

Historically PLC has mainly been used by the utility organizations for meter read-ing, load control and other smart grid applications, but has not really gained any big attention in the automotive industries. Some prototype constructions have been made, but PLC has not been embraced as a standard way of wiring.

1.2 Aim, Goals and Limitations

There is at least two ways to implement PLC in automotive use and these are done either by using the already existing wiring architecture with a single power wire and ground return or alternatively changing the wiring to a more PLC adapted architecture with an additional dedicated return conductor. Either way there will be a reduction of the amount of conductors by either one, two or three depending on the intended link of in-terest. To further reduce the amount of wire copper a type of bus architecture could also be implemented with the sensor power feed cables spliced into a single common network.

(25)

Chapter 1. Introduction 5

This thesis work is a part of a Volvo’s advanced engineering project that aims at inves-tigating PLC for powertrain sensor network over the course of two years. This thesis is thus only limited to 20 weeks and strives to demonstrate the feasibility of PLC with as simple means as possible. That is, whether it is possible to implement PLC on the present day wiring harness structure, as it is, by switching the physical media for in-formation carriers on top the power lines without adding filters or alternating the wire harness structure, and still securing reliable communication.

1.3 Objectives

The main objective is to investigate whether it is possible to secure reliable commu-nication on the present day wire harness structure, and to achieve this objective, the following points will be investigated.

• Theoretical background study on power line communication- Investiga-tion of how PLC work, what are the limitaInvestiga-tions, possibilities when implemented in a vehicle.

• A market survey on PLC

systems-– 1. Find out what sort of solutions are there in market. Are there any solutions that could cope with the harsh environment in a vehicle?

– 2. Determine the hardware needs to demonstrate PLC in a vehicle.

• Measure the channel characteristics- What is the impedance, gain and noise situation in a vehicle, is it possible to match the impedance and what should the signal to noise ratio be, and at what frequency should the PLC system operate. • Digital interface- What type of limitations are there in the digital interfaces that

can be connected on PLC? CAN, LIN, SENT etc.

• Bandwidth- How large bandwidth is required to implement PLC for powertrain sensors.

(26)

• Advantages and disadvantages- The advantages and disadvantages from a Volvos perspective with the PLC compared to the current solution what is regard-ing the economical aspects.

1.4 Outline of the Thesis

(27)

Part II

Theory

(28)

(29)

Chapter 2

Power Line Communication and

Theory Around it

2.1 Overview

Power line communication is simply what the name implies, namely using the existing power lines as a physical media for networking purposes. The power signals use signif-icantly lower frequencies and much higher voltages than what is used for information exchange, therefore the two signals can be superimposed on top of each other without interrupting the power waveform.

An illustration of how the waveform of information signal overlaid on top of an AC and DC power signal might look like is depicted in 2.1and 2.2

Figure 2.1: Communication over AC Figure 2.2: Communication over DC

(30)

and since this thesis work focuses on overlaying the powertrain sensor communication on top of the power-lines in a truck where the system voltage is 24V to 28V-DC, the DC-type of PLC will be of interest.

The intentional PLC system that should be constructed can principally be seen as in figure 2.3 where for instance a pressure sensor or any other sensor could be sending information in form of a CAN message from the electronic control unit (ECU) to the engine management system (EMS) or to the after treatment control module (ACM) that then processes the information. The digital CAN signal from the sensor must first be

Figure 2.3: A block diagram illustrating the PLC system.

modulated in a fashion that it complies with the channel characteristics of DC-lines. The modulated signal is then overlaid on the power line with the aid of a coupling cir-cuit that couples the signal through either by inductive or capacitive coupling, without putting the modem into danger from the line voltage. The receiving side then taps the high frequency signal from the DC-lines and performs a preliminary filtering before demodulating the signal into the original digital CAN message that was sent from the sensor.

(31)

Chapter 2. Power Line Communication Basics And Theory Around It 11

2.2 The Channel and a Channel Model

The word channel that is used in this section and in the rest of the thesis is a nomen-clature for the DC-lines that is ought to be used as a communication channel for the powertrain sensor network.

The DC-lines in a truck where never designed for communication purposes and that is why it is important to investigate the limitations of the DC-lines as a communication channel.

In conventional AC-grid applications the channel characteristics has shown to be af-fected several aspects [4], such as the effects of the line length, the length and number of branches and the effects of the time varying impedance of the line from connection and disconnections of loads which in turn alters the networks topology and could cause unwanted reflections due to improper impedance match [1] [25]. It is therefore the outermost importance to investigate if these similar types of effects cause issues when designing a PLC architecture for the automotive use and that is why the following sub-sections tries to highlight some of these issues.

The DC-lines in a vehicle has nevertheless a more deterministic topology compared to the AC-grid applications since the amount of loads are limited and they could be made more controllable which suggests that the matching of the impedances between the channel and the PLC modems, at least in theory would be simpler than in the AC-grid application.

2.2.1 Wire Harness

The wire harness constitutes the housing for several wires and builds up a quite complex topology as seen in figure 2.5 and how it is mounted on a engine is seen in figure

(32)

the suppling wire. This is all because there is usually a common ground connected to the chassis were the return conductors are connected and that is why there could be different lengths for the wires depending on physical location of the supplied load and ground point.

Figure 2.4: engine wire harness Figure 2.5: engine wire harness

2.2.2 Sensor Supply Lines as Transmission Lines

The length between the sensor ECU and the engine management system (EMS) can vary from few centimeters to few meters. and taking into consideration that a PLC modem can have a modulation frequency in the MHz range it becomes evident that the channel should be considered as a transmission line. This is because at higher frequencies where the wavelength is in comparable size with the length of the DC-lines there will be spatial variations in the voltages and currents along the line due to their respective wave prop-erties [3]. This is since there is a capacitance created by the deposition of charge due to the voltage difference between the two conductors and a inductance created by mutual and self wound magnetic field. These capacitances and inductances causes therefore a time delay for the voltages and currents applied to the DC-lines, since it takes a finite time to charge and energize the capacitance and inductance and likewise to decharge and deenergize them. These time delays cause the voltages and currents to propagate as waves along the lines since the changes applied to the voltage and current takes a finite time [17].

(33)

and electric fields are assumed to be perpendicular to each other and transverse to the direction of propagation. This is though not the only mode of propagation, since if the distances between the two conductors become too large in contrast to the wavelength, more complicated wave modes could be supported, and when the plus and ground con-ductor are not routed adjacent to each other the waves are propagating between the wire and chassis ground plane. Nevertheless by assuming that the propagation is in the TEM mode, then by definition, the Faraday’s and Ampere’s law will be simplified for the transverse field components of the electric, E, and magnetic, B fields into the following form as seen in equations 2.1 and 2.2

∇t× Et= 0 (2.1)

∇t× Bt= µJz (2.2)

which is the same as for the electrostatic case where the electric and magnetic fields are independent of each other. This implies that unique voltages and currents can be defined at any point in the transmission line and this assumption leads to a simplification of the calculations into a distributed-element circuit which is utilized in the appendix A, where the results are used in next subsection to describe various effects.

A general rule of thumb is that if the channel length is greater than one tenth of the sig-nal wavelength (λ/10) then the channel is ought to be considered as electrically long and to be treated as a transmission line [17]. The association between the signal frequency and wavelength is illustrated in equation 2.3

v

f = λ (2.3)

where v is the speed of the electromagnetic wave which depends on the medium between the two conductors (close to speed of light, c) and f is the frequency of the modulated signal.

(34)

of voltage and current occurs at every location where the impedance is not equal to characteristic impedance, Z0, of the line, and these locations are found at every load

and at every stub or branch of the channel where the impedance deviates from the Z0.

This could be illustrated for the DC-lines as seen in figure 2.6 where the impedances Z1 to Z4 represents the loads connected to the DC-lines with different length of line

and different impedances. All these different impedances and lengths affects the chan-nel characteristics for the PLC network and could be considered as a possible source of reflections.

Figure 2.6: Illustration of loads connected to the DC-lines that could cause reflections due to improper impedance matching.

Other aspects that could cause problems with the PLC network are capacitive or low impedance loads towards ground that could shunt the signal current towards ground instead of the receiving side of the PLC link. These low impedances could arise from the EMI (Electromagnetic Interference) filters situated in the sensor ECUs and from the impedance of the batteries.

2.2.3 Simple Transmission Line Model

(35)

into consideration, which would thus need the use of multi-conductor transmission line theory [16]. Furthermore the effects of the stubs and branches with variable loads should also be taken into consideration and the difficulty in taking these into considera-tion is that the distance between the lines and the ground plane is not uniform and the loads connected to the stubs can not be considered constant for all times and frequencies.

A way of showing how the channel affects the power transmission in higher frequen-cies is to study the input impedance, Zin, which in turn correlates with the amount

of power received at the receiving end, since the maximum power transfer is obtained when the input impedance, Zin, is equal to the internal impedance, Zg of the source [18].

The input impedance is the impedance seen by a transmitting PLC modem looking to-wards the load as it is depicted in figure 2.7 where the source and Zg can be seen as

the transmitting PLC modem and its internal impedance. The Z0 is the characteristic

Figure 2.7: Transmission line with the input impedance, Zgand load impedance ZL

impedance of the transmission line which is the ratio between the voltage and current waves propagating without any interference from reflected waves. The characteristic impedance depends only on the geometry of transmission line and is not affected by the length of the line, although the input impedance Zin does. ZL on the other hand is the

load or terminating impedance which is connected to the other end of the line and this could be thought as the impedance at receiving end in a PLC network.

The expression for the input impedance Zin can be seen in equation 2.4

Zin= Z0

ZL+ jZ0tan(βl)

Z0+ jZLtan(βl)

(2.4)

where the dependency is in the length of the line, l and in the wavenumber, β, which in turn has frequency dependency through equations 2.5 and 2.3.

β = 2π

(36)

Figure 2.8: How the input impedance changes with respect to frequency in a one meter long cable that is terminated with a 50Ω load. (For source code, see AppendixB)

The expression for the input impedance and the characteristic impedance, Z0, is derived

and shown in appendixA.

To illustrate the effect of input impedance, a one meter long power cable that is used for supplying N Ox sensors could be considered. This cable has two conductors with an

sur-face area of 1 mm2 which in turn corresponds to a radius of about 0.56 mm. The wires are approximately 2 mm apart from each other and has both insulating material (PVC) and air between them. The cable is thought to be fed by a source with an impedance of 50 Ω and terminated with an equal impedance.

From figure 2.8 it can be seen that the channel does not affect the input impedance for the lower frequencies, that is in this case frequencies below 1 MHz. For higher frequencies the channel reactance alternates between inductive and capacitive behavior depending on the load impedance, and the resistance peaks when the channel length is one quarter of the wavelength or an odd integer multiple of a quarter wavelength. The impedance becomes then purely resistive with the value of

Zin=

Z₀2 ZL

(37)

Figure 2.9: How the input impedance changes with respect to frequency in a one meter long cable that is terminated with the characteristic impedance. (For source

code, see AppendixB)

But on the other hand if the line would be terminated with an impedance equal to char-acteristic impedance, the effects seen in figure 2.8would be eliminated since there would be no reflections. To terminate communication lines with an impedance corresponding to the characteristic impedance is a common method and this is for instance found in the automotive industry with all the twisted wires used for CAN communication that are usually terminated with 120 Ω on both sides of the line. The effects on proper termination could be illustrated with this model in figure 2.9where the termination is 140 Ω and it can be seen that the frequency varying effects are disappeared. The proper termination is of course easier said than done since in the DC-lines the terminating impedance varies with the loads that are not adjustable.

(38)

The expressions for the load and source powers, PL and Pg are derived in appendix A

and the quota between them, as it is in equation 2.7is plotted in figure 2.10.

Power Ratio = PL Pg = 1 | Zin+ Zg |2 < Zin (Zin+ Zg)∗ (2.7)

From the figure 2.10 it can be seen that when the source impedance and the load impedance are the same as the characteristic impedance the power ratio is constant at 1/2 for all the frequencies, meaning that half of the power is dissipated at the source and the other half is dissipated at the load. This is an inevitable consequence if maxi-mum power is wanted to be transfered from the source to the load. For the case when the source and load impedances are 50 Ω the power ratio is depending on the input impedance, Zin and is varying with the frequency as a consequence of it. When input

impedance peaks at 402 Ω, so does the power ratio as well. The power ratio in this case reaches almost to 0.9 meaning that with higher load impedance there is better efficiency achieved but at the cost of the amount of power delivered. The question should also be asked whether the maximum power is the best practice for the PLC communication purposes or whether it would be better to just have high load impedance which could result in higher signal voltages for the PLC system.

(39)

Figure 2.10: How the ratio between the power delivered to the load and power de-livered by the generator varies with respect to frequency in a one meter long cable that has both 50 Ω and Z0for the internal and load impedance. (For source code, see

Appendix B)

2.3 The Coupling Circuit

The purpose of the coupling circuit is two-fold. Firstly it needs to protect the PLC modem from the higher voltages of the power lines and secondly it needs to provide a path for the signal voltage.

There are two main methods of coupling the signal voltage on top of the channel, and these are done either by capacitive or inductive coupling.

2.3.1 Inductive Coupling

In the inductive coupling a linked magnetic flux between a coil that is wound around a magnetic core and clamped around the DC-lines induces a voltage in agreement with Faraday’s law just as in a power transformer

E = −dΦ

(40)

where E indicates the induced voltage in the DC-powerlines caused by the interconnected time varying magnetic flux, φ. The benefits of using inductive coupling is that it provides an isolation for the signal source from the voltages of the DC-lines and that there is no need to mechanically connect the coupling circuit to the DC-lines since the signal wires and the DC-lines can be just wound around a ferrite core in order to complete the magnetic circuit for inductive connection. But since most of the PLC modems need to get their power from the DC-lines anyway, the ability of non mechanical connection becomes unnecessary since this connection must be done either way. Another drawback with inductive coupling is that the non-linear behavior of the B-H curve can cause distortion in the signal if the magnetic flux density becomes to large [20]. And also if the core becomes saturated by high currents on the DC-lines the inductions cease and no signal is induced on the DC-lines.

2.3.2 Capacitive Coupling

In capacitive coupling circuits the signal power is coupled through the displacement current through a capacitors, as seen in equation 2.9

Id= C

dv

dt (2.9)

where Id is the displacement current and C is the capacitance of the coupling capacitor

and v is the signal voltage sent by the PLC modem. The benefits of using the capacitive coupling is that the return path for the signal current is more controlled than in the inductive counterpart, since the capacitive coupler provides four connection terminals compared to two with the inductive. The downsides are that the capacitive coupling through a capacitor does not provide any galvanic isolation since there is a common ground and a mechanical connection is a must. The modulation frequency must also be chosen to be below the self resonance frequency of the capacitor which could occur due to parasitic inductances in the leads of the capacitor.

2.3.3 Common Mode and Differential Mode Signaling

(41)

In the differential mode, the modem is connected between two wires of the DC-lines and the signal voltage constitutes thereby a balanced configuration towards ground and thus the signal is only transmitted through differential mode. On the other hand when using common mode transmission the modem is connected between one of the wires and the ground, this configuration constitutes thereby a unbalanced configuration towards ground and both common mode and differential mode is excited for the signal transmission. The common mode coupling has shown yield up to 30 dB better coupling than using differential mode [15] although this would result in higher electromagnetic emissions [23].

Figure 2.11: balanced signaling Figure 2.12: Unbalanced signaling

2.4 Modulation Schemes

To overcome the nonlinear channel characteristics and to provide a compatible commu-nication between the nodes in PLC network the digital signal is usually modulated into a sinusoidal carrier signal using different schemes.

There are different schemes for modulating a digital signal on top of power lines and the most simplest, narrow band, forms are described in the following subsections. These simple modulation schemes are of particular interest since these offers relatively cheap solutions compared to other more complex modulation schemes such as spread spectrum techniques [9].

2.4.1 Amplitude Shift Keying

(42)

f and with an amplitude of A in the following way      S1 = A sin (2πf t) S0 = 0 (2.10)

where S1and S0represents digital one and zero respectively. A down side with amplitude

shift keying is the discontinuities from the transition from a one to zero (at least with on-off-keying) causes a unnecessary allocation of the bandwidth which could ultimately be a source of EMC issues.

2.4.2 Frequency Shift Keying

For the frequency shift keying the digital one and zero is converted to respective sinu-soidal signal with different carrier frequencies f1 and f2

     S1 = A sin (2πf1t) S0 = A sin (2πf2t) (2.11)

Where once again the S1 and S0 represents digital one and zero. It should be important

that frequencies are close to each other in the frequency domain to assure that the channel behaves similarly for both of the S1 and S0 because of the frequency depending

loads present in the channel.

2.4.3 Phase Shift Keying

With the phase shift keying both the digital one and zero are modulated into a sinusoidal signal with the same carrier frequency. To distinguish them a part, a phase shift, ϕ, is introduced as described in the equation below.

(43)

2.5 Electromagnetic Compatibility

Electromagnetic energy can be unintentionally be coupled from one device into another and affect the operation of that victimized device in a negative way. These types of elec-tromagnetic emissions are unwanted and are strived to be controlled. Elecelec-tromagnetic compatibility (EMC) is the branch that aims to ensure that these emissions are below a certain limits so that compatibility between different devices is hopefully ensured.

The way the electromagnetic energy can be coupled from a source to a victim is usually divided into different paths [24], such as conductive coupling, where there is a direct connection between the source and victim through cables, transmission line etc.,. Short distances between the source and victim can cause the energy to be coupled inductively or capacitively and finally the energy can also be radiated by the source and be picked up by the victim because of various antenna effects.

The PLC technology should not be overlooked what is regarding the EMC and the fol-lowing subsections will go through some thoughts regarding these aspects which could be a possible source of future EMC problems.

2.5.1 Radiated Emissions

It was shown in [7] that the radiated emissions were lower for twisted differential wires carrying a PLC carrier signal compared to wires carrying a digital CAN signal because of the lower spectral content in the PLC carrier signal. These measurements does not nevertheless take into the consideration how the DC-lines are routed actually in a vehicle where the asymmetry between the plus and ground wire could lead to conversion of the differential mode current into common-mode, which has been shown in to cause substantially more radiation, as described in the following equations [23]

(44)

where ˆEDM and ˆECM represent the maximum electrical field strength in V/m, caused

by differential mode and common-mode current respectively, in a two conductor system with a line separation of d and with a length of L and measured at a distance R meters away from the wires. From these equations it becomes clear that there is substantially more radiation caused by the common-mode current than by the differential mode cur-rent and the reason for this is that the magnetic fields from the common-mode curcur-rents does not cancel each other out as in the differential-mode current since the currents on the two wires are in phase. This does therefore lead to an increase in the radiated field because of the additive magnetic field [23].

Interesting is also that the radiation caused by the differential mode current depends on the area that is spanned by the two conductors, and considering the fact that the return conductor is not routed adjacent to the plus wire in the grounding scheme found i vehicle which could lead to severe radiation caused by the large area spanned by these wires.

2.5.2 Conducted Emissions

The signal voltage is directly coupled on top of DC-lines which would inevitably draw some amount of current corresponding to the impedance present. This current can then cause problems in other devices that are placed on the same power network either by direct coupling or by radiation as described earlier. The magnitude of the allowable current that is placed on the DC-lines is regulated by the CISPR 25.

2.5.3 Susceptibility

(45)

Part III

Implementation and

Measurements

(46)

(47)

Chapter 3

Choice of Hardware for the PLC

system

3.1 Requirements

To determine what type of hardware to choose for the PLC system that was intended to be constructed for the powertrain sensor network, some requirements needed to be taken into consideration, such as bandwidth, modulation frequency and the digital interface. The requirements for the hardware in this project is solely limited to the electrical as-pect and does not take into consideration engine vibrations and the extreme temperature conditions that could be present in vicinity of the engine.

The following subsections has the intention to give some insight into the thoughts and ideas regarding the above stated requirements.

3.1.1 Digital Interface

One of the main aspects of this project is to give incentives to switch the remaining sensors with analog interface into digital for more robust signaling and for better fault detection purposes. Since there are already sensors in place with the digital communi-cation protocol CAN (Control Area Network) that is a well established communicommuni-cation protocol within the automotive industry, this makes it into a strong candidate regarding

(48)

the choice of digital interface since it would also provide backward compatibility with the few sensors already in place. Another suitable digital communication protocols that could be of interest is the LIN (Local Interconnect Network) which is also a established communication protocol that could allow relatively cheap hardware nodes.

3.1.2 Bandwidth

To estimate the bandwidth needs for this sensor network a N Ox sensor that uses CAN

communication has been taken as reference. This sensor sends 8 bytes per every 10 ms which corresponds to a bandwidth allocation of 64bits every 10 ms which is equal to 6.4 kb/s. A coarse estimation is then to approximate that if all the 15 powertrain sensors are converted into CAN communication with approximately the same speed as the N Ox sensor and placed on the same network, then the bandwidth for the sensor

network becomes 96 kb/s. This thus means that the speed of the PLC modems could be chosen relatively slow for power saving purposes.

3.1.3 Modulation and the Carrier Frequency

The modulation frequency of the PLC modem should be chosen with regards to the channel characteristics of the DC-lines. Which will be investigate later on.

3.2 Products Chosen for the PLC

The market survey on PLC solutions for automotive use gave little results since most of the PLC solutions seemed to be either for meter reading purposes with too low band-width or for home networking purposes with too high bandband-width. Nevertheless two types of evaluation boards were chosen for testing purposes from a company called Ya-mar electronics ltd, which has its own little niche in automotive PLC systems.

(49)

Chapter 3. Choice of Hardware for the PLC system 29

Figure 3.1: SIG60 Figure 3.2: DCAN250

of the DCAN250 is at 5 MHz where as the SIG60 has a variety of selectable frequencies ranging from 1.75 MHz to 13 MHz. The maximum baud rate of the DCAN250 is at 250 kbps and for the SIG60 it is at 115.2 kbps. For more information regarding these components the reader is referred to Appendix C.

3.3 Functional Test of DCAN250 and SIG60

The function of the DCAN250 and SIG60 were tested to see how to the signal was modulated and transmitted through the DC-lines. The measurements shown below was made to the DCAN250 and SIG60.

3.3.1 DCAN250

How the signal is routed through the PLC link is presented in figure 3.3 where an arbitrary CAN signal is fed into the DCAN250 thorough CANanalyzer, and the signal is measured using an oscilloscope (Tektronix TDS2012C) in various positions. In figure (a) of figure 3.3, both the inserted digital CAN signal and the sinusoidal PLC carrier signal is presented.

(50)

shown in figure (c) in 3.3. So totally there is a latency of about 770 µs introduced for the CAN signal when passing through the PLC link. This is a rather high latency and it should be taken into consideration. The ECUs e.g., has a general requirement that the latency should not be higher than 50 ms. It should nevertheless be mentioned that these latency measurements where made on a single DCAN250 modem by switching the transmission direction of the CAN message with CANanalyzer software.

(a) Both the CAN signal and the modulated PLC signal. (b) Latency from modulation of the CAN signal to a PLC signal

(c) Latency caused from demodulating PLC signal to CAN signal

(d) Spectral content of the of PLC signal.

(51)

Chapter 3. Choice of Hardware for the PLC system 31

The spectral content of the PLC signal is presented in figure (d ) in 3.3which is measured using the FFT (Fast Fourier Transform) algorithm of the oscilloscope using a flattop window function. From the image the 5 MHz carrier signal is clearly visible and also the third and 5:th odd harmonics at 15 and 25 MHz which are probably caused by the crystal oscillator in the modem.

3.3.2 SIG60

For the SIG60 the signal is sent from one computer to another using the SIG60 devices in the link between. Through the link, the letter ’A’ is sent in ASCII code continuously and the way they are modulated onto the DC-lines is illustrated in figure 3.4.

The spectral content of the SIG60, as seen in figure 3.5, was measured using the FFT function of the oscilloscope utilizing a Hanning window function, and from the spectrum analysis the 5.5 MHz carrier frequency and the three first odd harmonics is shown.

Figure 3.4: To the left: several ’A’ letters sent along PLC link using SIG60. To the right: One ’letter’ signal enhanced.

(52)

(53)

Chapter 4

Channel Characteristics and PLC

Test Measurements

4.1 Motivation for the Measurements

To be able to determine the technical requirements for PLC modems suitable for pow-ertrain applications, some measurable properties of the channel could be of interest, namely the noise level, impedance and gain. These measurable quantities could have a significant effect on the choice of PLC modem regarding the needed signal amplitude, choice of input impedance and modulation frequency in order to gain as high signal to noise ratio as possible.

• Noise. It is important to investigate the noise level present on the power lines for different frequencies. This is because the modulation frequency should be chosen at a frequency where the noise level is at its lowest and from the noise level it is possible to estimated how high the signal to noise ratio should be.

• Characteristic Impedance and the Line Impedance How does the line impedance vary during different operating modes and what is the characteristic impedance of the power lines. With these parameters it could be decided whether it is possible to match the impedance of the PLC modem towards the line.

(54)

• Forward Gain, (S21) The amplitude for the PLC signal for different frequencies

can be observed and from the results hopefully a carrier frequency can be chosen where the attenuation is at the lowest.

4.2 Device Under Test, Link of Interest and Break Out

Connections

The device to test the PLC on was a Volvo FMX truck with an Euro 6 engine and TEA2+ electrical platform (Internal Reference: FM253). The truck was tested upon between the days 21 to 23/01/2013 and the link between after treatment control module (ACM) and the post catalytic NOx sensor was chosen because this link has rather long

wires when considering powertrain sensors. This link is around two meters and is one of the longest, if not the longest powetrain sensor link, and if the length of the wires would be a fundamental cause of problems for the PLC, it would certainly manifest in this link. Furthermore this link provides a bit more realistic picture for the PLC since the power to the NOx sensor is routed through the ACM which is also where the NOx

sensor signal is routed. Also this link provides relatively easy access for measurements. A schematic illustration of the link between the ACM and the NOx sensor is given in

figure 4.1. It should be noted that this figure does not exactly comply with the truck under test but merely demonstrates the distances and positions of the ACM and the NOx sensors.

The NOx sensor receives its supply voltage from the ACM (pin 49) which gets its own

voltage directly from the battery. This voltage is filtered at ACM by use of some series inductors and parallel capacitors before received at the NOx sensor. The return path

from the NOx sensor is routed to the common ground (YA:3C XS373) of this particular

(55)

Chapter 4. Channel Characteristics and PLC Test Measurements 35

Figure 4.1: Link between the ACM and NOx sensors which shows the approximate

(56)

Figure 4.2: Break out connection for NOx

Figure 4.3: Break out box for the

ACM

4.3 Measurement Setup

To be able to make measurements at this link, a pre-made break out box as seen in figure

4.3was connected to the ACM. This break out box provided break out connections for all the wires running in to the ACM so that measurements were made possible. The wires from the break out box to the ACM where approximately one meter which also increased the distance of the link by the same amount and this could have some impact on the channel characteristics.

For the NOx sensor a self-made break out connection had to be constructed, as seen

in figure 4.2. This break out device provided all the necessary connections for the mea-surements to be made. That is, it provided the connection points for the PLC modem and had a coaxial cable soldered onto it to make s-parameter and noise measurements possible.

The measurement setup used for all the measurements during this test can concep-tually be seen in figure 4.4. For the s-parameter and noise measurements the same instrument was used (Rhode & Schwarz FSH8 (internal reference: EMC 60462))) which could be used as a scalar network analyzer as well, and since it could only measure the scalar values the impedance measurements were thus omitted.

(57)

Figure 4.4: Measurement setup used for all the measurements

center core and the plus wire of the sensor (see figure 4.4for illustration). At the Nox

sensor side these capacitors where soldered on to the self made break out device and at ACM side the coaxial cable with capacitors where connected to its break out box with banana contacts.

4.4 S-parameter Measurements to Estimate the Transfer

Function

The scattering parameters or more simply the s-parameters are dimensionless parame-ters representing the quotients of reflected and forward moving waves [4]. For a two port system, there are four different s parameters which are more specifically the s11,

s22, s12 and s21. The s11 parameter represents the reflection coefficient that occurs at

port 1 while there are no reflections occurring at port 2. Similarly the s22 represents

the reflection coefficient that occurs at port 2 while there are no reflections occurring at port 1. The parameters of particular interest regarding channel characteristics measure-ments used in this thesis are nevertheless the s21and s12 parameters. These parameters

(58)

Figure 4.5: s21 measurements made between the frequencies 100 kHz and 1 GHz

which shows a clear low pass characteristics with severe resonant attenuations.

made here below is that measurements here were made with the scalar magnitude in dB, compared to the complex magnitude made in the references.

To start with the s-parameter measurements a s21 sweep was performed while the truck

was completely off. This sweep was made between the frequencies 100 kHz to 1 GHz and the results are shown in figure 4.5. The results show that there is a clear low pass characteristics and severe attenuation presented on this link, most likely due to the capacitive loads present at the supply voltage filter located at the ACM and probably at the EMI filter located at the NOx sensor ECU. The link was also extremely frequency

selective regarding the resonant attenuations which are seen as notches in figure 4.5

and these notches are described in [25] to be caused by several reasons such as the number of branches, length of the branches and the load reactances. In [4] it is also described that bonding of the return wire and ground creates a significant resonant at-tenuation as well. It is difficult to say what is caused by what in this case, but it could be assumed that all these factors has a certain contribution to what is seen in figure 4.5.

To see whether there was a difference in the link between the ACM and NOx sensor

(59)

Figure 4.6: Reciprocity of the link between the frequencies 1 to 100 MHz. Although these measurements shows close similarities the measurements made with the PLC

modems contradicts these findings.

measurements where made by comparing the s12 and s21 parameters between the

fre-quencies 1 MHz to 100 MHz and the results show a rather good reciprocity of the link. Although the reciprocity was shown, this did not reflect the reality of the behavior of the link when it was tested with the DCAN250 PLC modems, because when tested with the PLC modems it was shown that the link performance was better in the NOx

to ACM direction compared to the ACM to NOx (more of this in section PLC test).

(60)

Figure 4.7: s12-parameter measured between 1 MHz to 100 MHz during different

conditions: Truck being off, at pre-state mode, with engine idling and with a separate return conductor. It is shown that there is no significant difference in the modes besides

with the separate return conductor.

These measurements on reciprocity nevertheless show that the forward transmission coefficients are not sufficient to describe channel symmetry, and measurements on the impedances of the channel should probably complement these type of measurements.

The s12 parameters where further tested and compared with different conditions of

operations namely with the truck being off (key in far left position), the truck in a pre-running state (key position before ignition) and with the engine idling, and furthermore, the s12 parameters where tested with a separate return conductor connected from the

NOx sensor to the ACM (more of this in section PLC test).

Figure 4.7 shows the results of these measurements and it shown that there is no significant difference between the modes of operation. Nevertheless it should be noted that the difference in impedance in these different modes is not shown and thus the measurements of the s12 parameters does not give the complete picture of the channel.

To see how sensitive the channel was for connection of different loads and branches the DCAN250 PLC modem was connected to the NOxsensor side and the s12was measured.

(61)

Figure 4.8: s12 measurements made with the DCAN250 connected to the NOx side

between 1 MHz to 100 MHz which shows that connected loads has a significant impact on the channel characteristics.

the results, as seen in figure 4.8 clearly shows that the connection of a load at the link causes severe notches at the measurement results compared to the case without.

4.5 Noise Level Measurements

It was also of interest to see where the noise level was at these different modes of operation (truck off, pre-running state, and engine idling). The noise level measurements where made at the ACM side of the link and the measurements where made between the frequencies 1 to 100 MHz. At the input port of the spectrum analyzer a 6 dB power attenuator was connected for protecting purposes. The findings of these measurements are found in figure 4.9 where it is shown that some narrow band noise is generated for the frequencies below 60 MHz and at the higher frequencies the overall level is increasing, while the truck was at pre-running state and when the engine was idling.

(62)

Figure 4.9: Noise level measurements made between the frequencies 1 MHz to 100 MHz where there it is shown that there is no significant different between the noise

level while the truck was in pre-state mode and while the engine was idling.

4.6 PLC Test

The PLC at this link was tested with the DCAN250 modems. The modem at the ACM side was connected between the supply feed cable of the NOx sensors (pin 49 at the

ACM) and ground (pin 58 at the ACM). The other modem was connected between the supply feed cables of the NOxsensor located after the catalytic converter (see figure 4.4

for sketch).

(63)

Figure 4.10: CANanalyzer screenshot

could be listened, and statistics of the sent and received messages could be recorded.

During the measurements, abbreviations were made to the different PLC modems to keep simpler records, and this is why the same abbreviations will be used for the differ-ent modems during the rest of this section. The modem located at the NOx sensor side

was abbreviated to CAN1 for the simple reason that it was connected to channel one of the CAN interface. Similarly the PLC modem at the ACM side was abbreviated to CAN2 because it was connected to channel two.

The measurements were once again done with different modes of operations, that is with the truck at pre-running state and with the engine idling. The measurements where the truck was completely off was omitted due to the simple fact that had to do with the power supply of the PLC modems.

(64)

and when either of these amounts surpassed 128 errors the CAN node was changed into an erroneous state where its transmitting capabilities were inhibited. That is why the measurements were conducted during the time till the CAN node entered this erroneous state whereafter the measurements immediately were stopped. For the cases where the erroneous state was not reached, the measurements were stopped after an arbitrary time.

The results from two cases where the CAN communication was tested with the en-gine idling, and with the transmission direction going from CAN2 to CAN1 is presented at table 4.1. The table shows the extended data messages received and transmitted at channel one and two and whether the CAN nodes had entered an erroneous state.

Ext. Data [total] CAN1 Ext. Data [total] CAN2 % Received Erroneous state

Test 1 2309 3444 67% yes

Test 2 8791 11708 75% yes

Table 4.1: PLC test results engine idling CAN2 to CAN1

The tests where proceeded with having the truck at the pre-running state and the same quantities where measured with the same manner as before. The results from these measurements are presented at table 4.2.

Test 1 5866 9461 62% yes

Test 2 8509 13232 64% yes

Table 4.2: PLC test results truck in pre-running state, CAN2 to CAN1

The same procedure was also done with changing the transmission direction so that CAN1 was now transmitting the same message to CAN2. The results from the test with the engine idling is presented in table 4.3.

Test 1 1505 1098 72% yes

Test 2 13118 12562 95% yes

(65)

And once again the measurements were made with the truck being at the pre-running state but the communication direction shifted and the results tabulated in table 4.4.

Test 1 13999 13403 95% no

Test 2 13807 13556 98% no

Table 4.4: PLC test results with the truck being at the pre-running state, CAN1 to CAN2

Before summarizing the results, the same type of tests where also performed after re-placing the return conductor of the the NOxsensor by a wire with a wire gauge of 1 mm2

and with an approximate length of 3 m. This wire was then connected at the ground node of the ACM (pin 58) which is the same place where the CAN2 PLC modem return wire is connected. A sketch of this connection is presented at 4.4.

Once again the measurements where made with the direction of the transmission going from CAN2 to CAN 1 and with the engine idling and with the separate return conductor and the results presented at table 4.5.

Test 1 10160 10400 97% yes

Test 2 19101 19417 98% yes

Table 4.5: PLC test results with the engine idling and with a separate return con-ductor, CAN2 to CAN1

Now the results with the truck at the pre-running state and with same communication direction as before (table 4.6).

Test 1 12580 13009 96% yes

Test 2 6848 7590 90% yes

(66)

Figure 4.11: CAN communication results from CAN1 to CAN2

Figure 4.12: CAN communication

results from CAN2 to CAN1

And finally the measurement results with the communication direction changed to CAN1 to CAN2. And to begin with, the results with the engine idling is presented in table

4.7.

Test 1 23522 23522 100% no

Test 2 70442 70442 100% no

Table 4.7: PLC test results while the engine is idling and with a separate return conductor, CAN1 to CAN2

And to end with, the measurements with the truck at pre-running mode is presented in table 4.8.

Test 1 22852 22852 100% no

Test 2 20451 2450 100% no

Table 4.8: PLC test results while truck is at pre-running state and with a separate return conductor, CAN1 to CAN2

(67)

From the results seen in these graphs, there is a clear difference whether the signal is transmitted from the CAN1 side (NOx side) or at the CAN2 side (ACM side) of

the link under observation. This is therefore contradicting the reciprocity findings as shown in figure 4.6 which could have to do with the different source impedances for the PLC modems and measurement devices and that the transfer function therefore be-haves largely different because of it. Another solution to this could be that the voltage transfer function simply is not sufficient to characterize the channel completely and that the loads connected to the channel do not provide the same frequency response in both direction. It should also be pointed out that the eventuality of any Monday specimens among the PLC modems where also tested by switching places of the modems to rule out the possibility of any difference in the PLC modems.

The connection of a different return path had a surprisingly positive effect regarding the data transmission. As it is compared with its regular return path, the change in the direction CAN2 to CAN1 was around 23 percentage points (pp) better while the engine was idling, and 34 pp while the truck was in a pre-running state. In the other direction, from CAN1 to CAN2, the difference was not that significant where there was a slight increase of around 4pp while engine was idling and 2 pp while the truck was pre-running, but although the change in percentage was not that significant, the main improvement was that in the transmission direction CAN1 to CAN2, the nodes did not enter an erroneous state and also in the direction CAN2 to CAN1 there was a significant increase in the number of data messages that was sent before these nodes entered an erroneous state. More specifically the increase was around 66 % while engine was idling and 53 % while truck was in the pre-running state.

(68)

(69)

Chapter 5

EMC Measurements

To see whether the superimposed signal on top of the power lines would be a potential EMC issue in the future, the radiated emissions were tested in an anechoic chamber. The radiated emissions were tested according to the volvo standard: std 515-0003 ver-sion 4 for component tests, which in turn refers to CISPR 25, 3:d edition. This test were carried out by having two wires with a 1 mm2 cross sectional area and with a length of 1.5m placed above a ground plane in accordance with the standard. The DCAN250 modems where placed on both sides of the wires and the wires were fed with 15 V between them through a LISN (Line Impedance Stabilization Network) to provide a precise impedance towards the power supply. To get the PLC modems function in the anechoic chamber, the same type of setup that was used in the previous measurements were used, namely using the CANanalyzer to generate messages and transferring them through a CAN interface into the modems. The only difference during this test was that to avoid any other sources of disturbances besides the 1.5 m long wire in the anechoic chamber, the CAN messages where converted into optical signals using a CAN-to-optical transducers devices which is depicted in figure 5.1. The whole measurement setup (for lower frequencies) is shown in figure 5.2.

During this test the radiation that had a polarization vertical to the lines were measured over the frequencies ranging from 150 kHz to 3 GHz and the average and peak electro-magnetic (EM) field strength was measured and the results are shown in figure 5.3and

5.4.

(70)

Figure 5.1: CAN-to-optical trans-ducer and DCAN250

Figure 5.2: The test setup for EMC measurements

(71)

Chapter 5. EMC Measurements 51

Figure 5.3: Average radiated EM-field with vertical polarization from a 1.5 m long cable. The orange solid line marks limits for the average EM-field.

(72)

It is difficult to draw a single conclusion of what is the main cause of these radiations just from these measurements. But what is known is that the radiation is a consequence of the acceleration and deacceleration of charges [2] which would imply that the source of the radiation is the time varying current, and since the impedance of the line is quite low, the modems needs to compensate in current to keep the signal voltage level at its nominal value. This would be a posible cause of increase in the differential mode radia-tion which is proporradia-tional to the magnitude of the current and the loop area constituted by the two wires. Another or complementary explanation would be that the parasitic capacitances shunts the current towards ground, causing common-mode currents which are prone to cause more severe radiation due to the non canceling magnetic fields [17].

(73)

Part IV

Economical Aspects

(74)

(75)

Chapter 6

Economical Aspects

It is difficult to estimate what the impact of implementing PLC for the powertrain sensor network would be regarding all the economical aspects that this technology brings out. It is although the aim of this chapter to bring forward some of these aspects, if not by precise monetary calculations, then by some more hypothetical estimations and discussions. Thus the following sections tries to estimate and bring forth some of these economical aspects.

6.1 PLC Cost Savings for NO

x

Sensors at EATS-Wire

Har-ness

What is known is that sensors with the PLC technology, in theory, does not need any-thing else besides the power supply cables for transmission media. This would imply that there is no need for any sensor to have more than two wires, or one and a half, if single point grounding is used and the the return conductor to the grounding point would be shorter than the supply line. This would thus reduce the amount of the wires present in a vehicle drastically, but what really is the monetary gain in these reductions will here below be investigated for the EATS-harness (Engine After Treatment System) which is a wire harness connecting the NOx sensors into the the chassis harness which

is an another wire harness that prolongs the connection so that the connection between the NOx sensors and the ACM are made possible. This chassis harness is found and

(76)

hanging.

The EATS-harness holds housing i.a., for the twisted CAN wires and supply wires for the NOx sensors, and to start with the estimations of what kind of savings this PLC

technology would bring forth, the cost analysis done here is limited to a 21426747 EATS-harness which is the most common type of EATS-EATS-harness constructed. So to start with the savings for this harness the first thing to be removed are the twisted CAN wires. This would lead to reduction of the following wires with the corresponding lengths, which are listed in table 6.1.

Specific wire to be removed Length of the wire [mm]

EIC:11 to XE3 120 EIC:10 to XE3 190 EIC:8 to XE2 120 EIC:7 to XE3 190 B96B:1 to XE3 775 B96B:2 to XE2 825 B96:1 to XE3 481 B96:2 to XE2 551 TOTAL 3252

Table 6.1: The wires that can be removed from EATS-harness if PLC is introduced.

To further reduce the cost of this harness the connectors to the NOx sensors can be

(77)

Chapter 6. Economical Aspects 57

(Tyco electronics 970772) could also be removed. And finally, there are two splices on the communication wires which would also imply a small cost saving when they are removed.

All these changes listed above where presented to LEONI, a company that supplies wires and cables, and their estimation was that the cost savings would be approximately 1 e with also taken into consideration the labor time from twisting of the wires and routing. This is actually a quite substantial saving considering the size of the harness and it should also be pointed out that there are somewhat 60000 examples of these type of harnesses made yearly. Since these calculations are only made to the EATS-harness there is a need to do these similar type of calculations in all the harnesses affected by the PLC in order to get a complete picture of what the wiring savings would be.

6.2 Sensor Price Increase

The PLC does not come without a price. The sensors and the signal receiving side (ACM, EMS etc.) would be needed to be equipped with PLC modems which does increase the price of them. If it is assumed that there is a high competition on the PLC equipped sensors in the market so that the sensor price only increases with the marginal cost of the inserted PLC electronics, then this increase in price could be estimated if they are for instance equipped with the SIG60 modems from Yamar electronics, which has a price of 3.57 e for ordered quantities higher than 1000. Then the price of the sensors would also increases with this price if the other aspects such as coupling circuits and passive filters are assumed to be neglectable small. By considering the link with the NOx sensors again, there would thus be a increase in the price of the ACM and

the sensors with this 3.57 e. Of course other possibilities are possible such as that the closely located sensors could share a PLC modem to press down the price of the sensors. But to sum up there will be a price increase for all sensors and signal receiving devices.

Automotive Power Line Communication: A New Wiring Topology for Powertrain Sensor Network