CHAPTER 8

Communications Systems

iiiiiii~'iiiiiiiiI
INTRODUCTION 189

~!ii~i 2 NETWORK BASICS 191

iiiiiii
!iiii!!ii!ii!iiiii 3 SIMPLE FIELDBUS SYSTEMS 196

'~'~'~='~=~' 4 FIELDBUS SYSTEMS 197

INTRODUCTION

In a modem digital drive, adjustments such as PID gains usually involve multiple drives working in unison with each
and acceleration ramp times are software settings rather than other. A field engineer may be faced with the daunting task
the potentiometers and DIP switches found in their ana- of setting up the parameters for thirty drives, each one
logue predecessors. To permit adjustment of these software having over four-hundred parameters.
parameters, most drives provide a human-machine interface
(HMI) consisting of an onboard display and buttons. Users To address this challenge, most modem digital drives have a
manipulate the buttons to scroll to the parameter of interest built-in communications port using the RS-232 or RS-485
and to alter its value. The problem is that many sophisticated serial standard. There are two principal reasons for the
drives may have several hundred parameters and the built-in inclusion of a communications port on a digital drive: set up
HMI system can be cumbersome to use. Compounding this of the drive's parameters and real-time control of a number
problem is the fact that modem automation applications of drives in an automation application.
190 INTRODUCTION: Drive Set Up

Figure 8.1 Screen of typical drive set-up program

PLC drives

~ ~ :~::~ ~:~i~ ~

IIIIIIII lit referenCerun

reference
run
sT6
reverse
reference
run
stop__
reverse
direct wiring of signals

Figure 8.2 Traditional discrete interconnection wiring of a drive system

DRIVE SET UP
which permits a single drive to be connected; if an RS-485
Many drive manufacturers provide WindowsTM-based set up converter is used then multiple drives can be connected
and maintenance programs to acquire and alter the drive's through a multidrop cable. Control Techniques' drives can
parameters. These configuration programs can operate on a be configured and maintained by PC applications such as
laptop computer and attach to the drive via the serial com- Unisoft (for the Unidrive product range) which supports
munications port. The standard PC COM port uses RS-232 single and multidrop connections (Figure 8.1).
 Chapter 8.2 191

PLC drives

serial network

Figure 8.3 Serial network interconnection of a drive system

DRIVE CONTROL SUMMARY

Many applications require a number of drives to be inte- In conclusion, simple serial communication portals built
grated with I/O and PLC devices. The traditional approach is into drives greatly enhance the task of set up and maintenance
to use discrete wiring for the interconnections (Figure 8.2). of the drive's adjustment parameters. For real-time control of
the drive in an industrial process, a fieldbus offers the
Replacing the unwieldy wiring looms with a single digital
advantages of higher speed, simplified installation and
serial network is an attractive alternative, as it will result in
enhanced reliability. The benefits in replacing discrete
a dramatic reduction in the wiring and cost of installation
interconnections with a fieldbus are summarised below:
(Figure 8.3). Provided that installation guidelines are met
(i.e. maximum number nodes, trunk length and correct ter- • wiring costs are dramatically reduced especially if dis-
minations etc.) reliability will be improved simply due to tributed I/O is utilised
the reduced number of connections. The data remains in the • data remains in the digital domain and most analogue
digital domain with many analogue components and their components can be eliminated; no conversion errors,
associated conversion, repeatability and drift errors being no repeatability and drift errors and substantial cost
eliminated. The flexibility offered by a serial network also
savings
allows applications to be configured for specific end user • provided that the communication system is suitably
needs without adding extra cable connections. Internal robust the overall system becomes less susceptible to
parameters of remote nodes can easily be accessed for EM noise and ground-loop problems
remote supervision and data logging. • remote supervision and set up of the drives over the
The speed of standard PC serial communications may be network
adequate for set up and configuration activities, but is • if the selected factory network is one that has wide
usually too slow for this purpose. Consequently, a number of industry support, many sensors, actuators, HMI systems
high-speed industrial networks have evolved to meet this and controllers can be directly connected to it without
need. They are generally known as fieldbus systems. special interfacing.

2 NETWORK BASICS

Communications networks have so much in common that it is used as a framework for organising the various data com-
has become standard practice to relate their features and munications functions occurring between disparate devices
design elements to an internationally agreed model. The which communicate. The complete OSI model defines seven
Open Systems Interconnection (OSI) Reference Model, component parts or layers; however only three of these layers
developed by the International Standards Organisation (ISO), are usually developed for industrial communications.
192 NETWORKBASICS

Fieldbus systems often include special features related to required for the nodes to be connected in a logical bus
device interoperability and the real-time nature of industrial structure.
automation applications. The device profile defines device
interoperability features such as electronic data sheets which Interface Circuits
allow devices from different manufacturers to interoperate
without complex configuration or custom software. Cyclic All PCs support the RS-232 data transmission system. The
data is network data that bypasses parts of the software for signal appears as a single-ended voltage with reference to a
efficiency. signal ground. The voltage swing of the RS-232 circuit
shown in Figure 8.5 is -+-12 volts ( - 1 2 V for logic one,
+ 12 V for logic zero). When not transmitting the signal is
PHYSICAL LAYER held at logic one ( - 1 2 V).

Starting at the lowest layer in the OSI model (Figure 8.4), the Observing that the logic zero/one detection thresholds are
physical layer is concerned with the actual transmission of typically +3 and - 3 volts in most implementations, it is
raw bits. In a factory network, analogue quantities, switches, clear that a modest spike could trick the receiver into making
command codes and textual data are all converted into a false bit determination, as shown below in Figure 8.6. For
numeric information and transmitted as a stream of binary this reason, RS-232 signalling circuits are almost never used
bits from the source node to the destination node. Typically, in modern factory communications systems.
the bit stream is logically grouped into octets (or bytes). Adoption of differential signalling solves most of the
The following sections describe the key facets of the phy- problems inherent in single-ended RS-232 communica-
sical layer. tions. In the RS-485 standard, two conductors are used to
represent the bit: one carrying the original bit and the other
Network Cables and Connectors carrying its logical inverse. The differential receiver at
the receiving end subtracts the two signals to recover the
The physical layer requires a transmission medium for the original data. Any noise induced into one conductor is
data signals to flow. The principal transmission medium of induced in the other conductor and the subtraction operation
factory networks today is either copper cable or fibre-optic will thus cancel out the common-mode noise. Note: the
cable. signal levels are 0 V and 5 V.
Twisted-pair copper cable is preferred for cost and ease of RS-485 also supports operation with multiple nodes on the
installation. Fibre-optic cables are less susceptible to EM same cable. When a transmitter is not actually transmitting it
fields and offer higher bandwidths; however they are more is disabled and presents a high impedance to the bus.
expensive and difficult to install. Also, an expensive hub is Typically, all receivers are enabled and see every message,
but the protocol allows them to discard messages not
intended for them.

Data Encoding
The differential signalling described in the previous section
I~~................~Ye~i
layer .7:i..~i..i.application
.........~.......... layer
~.:.~..:.~..i~~iP~P:,l:,i~~~i
~i~~~i ....~.i.....~..........~...~...~...........
...~..i~..ila..ye: cyclic
data
allows us to reliably send a single bit down a network cable.
Since the network is typically used for numeric data or
characters, multiple bits are used to convey the information.
layer 2: data link layer These multiple bits must be encoded either asynchronously
or synchronously.
In asynchronous encoding it is up to the receiver to properly
sample and detect the multiple bits. The standard PC COM
Figure 8.4 Open Systems Interconnection port uses a method of encoding called NRZ (non return to
zero). In NRZ encoding, the voltage level determines the bit

RS-232 + 12 volts
__~~/interface >~/ ] ~ logic zero
................ + 3 volts
)> >> Ovolts
m m

m n

3 volts

idle
\ logic one

- 12volts
Figure 8.5 RS-232 interface
 Chapter 8.2 193

................... ................... + 25 volts

logic zero . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . + 3 volts

0 volts
~ detection!

'°°'c°ne 1"l"--t
......... v°'*s

........................................ 25 volts

Figure 8.6 RS-232- effect of noise

+5
,,, >> Rx ~tr data
ansmitted
data data

!]'Tx >> >> ~x

+s
STx' I-I

m B

+5

note: the spike has been cancelled by the differential

amplifier
kS )
+5

f 0a*areceived

Figure 8. 7 RS-485

value (one or zero). In Figure 8.8, the character Hex 1C is Manchester encoding, there is always a transition in the
transmitted. The receiving electronics must detect the middle of a bit period. Logical zero is a downward transition
change from line idle state to the start of a bit pattern (called and logic one is an upward transition. Since there is always a
the start bit). Once the start bit has been detected, each data transition, a phased-lock loop circuit can be used to extract
bit must be sampled in the middle of its respective bit period the clock signal. This makes it easy for the receiver circuit to
(Figure 8.9). This assumes that the receiver has its own sample and detect the bits.
sample clock to do this and that the incoming bit rate (baud
rate) is known.
The disadvantages of NRZ encoding is that the timing of the Network Topology
sampling is independent of the transmitted signal. Also, the
start bit and stop bit carry no information and thus waste The topology of a network describes how the nodes are con-
throughput. nected together. The main topologies are the bus topology
where all nodes connect together onto a common medium,
Another encoding method called NRZI (nonreturn to zero
and a ring topology where nodes are interconnected in
inverted) uses transitions to determine the bit values (zero is
a unidirectional loop. Some networks are wired in a star
no transitions, one is a transition).
topology which requires the use of a multioutput repeater
Finally, the Manchester encoding system allows the called a hub. Ethernet 10baseT is an example of this
clock signal to be recovered from the transmitted data. In approach.
194 NETWORKBASICS:Data-link Layer

logic 1 0 0 1 1 1 0 0 0

start bit bit bit bit bit bit bit bit stop
bit 0 1 2 3 4 5 6 7 bit

Figure 8.8 Transmission of Hex 1C

0 0 1 0 1 1 0

Figure 8.9 Receiving Hex 1C

DATA-LINK LAYER CARRIER-SENSEMULTIPLEACCESS/COLLISION

DETECT(CSMA/CD)
The data-link layer is responsible for encapsulating the
digital information into message frames and for the reliable A node waits until the bus is idle and then transmits its
transfer of frames over the network. message. While transmitting, the node senses its own
transmission to determine whether a collision with another
node occurs. If a collision is detected then several schemes
Framing
can be employed to arbitrate. Ethernet specifies that the node
An example of a data-link frame is shown in Figure 8.10. jams the network and then backs off for a random time
interval before trying again. As network loading increases
Data Model the probability of collisions increases and the network
ceases to operate efficiently. This stochastic behaviour is not
All industrial fieldbuses work by connecting devices onto suitable for the majority of automation applications.
a shared medium and multiplexing data onto this medium
in a serial fashion. The data model describes how messages CAN is also CSMA but the collision is resolved using
or data are routed and identified on the network. bitwise arbitration according to the priority of the mes-
sage. Significantly, this arbitration is accomplished on the
SOURCE-DESTINATION MODEL physical layer itself using dominant bit signalling and
does not result in wasted bandwidth. This signalling does
Messages are identified by a single node destination address. have a bit rate and trunk length limitation of 4 0 m at
Most source/destination protocols also have the possibility 1 Mbit/s due to the finite propagation speed of the signal
for global addressing so that all nodes receive the message. down the wire.
However, it is not possible to select a group of nodes to
CAN is deterministic for the highest priority message with
receive the message (called multicast).
a worst-case delay time of 130 gs (time for the maximum
eight-byte message at 1Mbit/s). Performance can be
PRODUCER-CONSUMERMODEL
improved further by using the hardware transmission time-
A message produced by a node is identified by its content stamp to measure the delay and then transmitting this value to
(i.e. data identification) rather than a specific node desti- the slaves so that they can compensate for the jitter.
nation. Any node may consume this message if it detects
that the data is required. Clearly, this is very powerful and TOKEN RING
makes best use of available bandwidth but does require a
complex configuration phase when data identifications A node may only transmit a message if it is in possession of
are allocated. CAN, FIP and ControlNet can successfully the token. Once the node has sent the message it must
operate a producer-consumer model although CAN only relinquish the token to its neighbour. This fair-share scheme
provides for a maximum of 2047 data objects. provides a very flexible protocol for peer-to-peer commu-
nication without the need for a master or arbiter node.
Media Access Control However, the worst-case time window for a node to receive
the token and transmit a message is large and occurs when, in
The media access control protocol defines how access to the one token pass, every node transmits a message of maximum
shared medium is arbitrated so that reliable data exchange length. ARCNET is the best example of an industrial net-
can occur. work which uses token ring.
 C h a p t e r 8.2 195

i/ preamble data
I this is the data we want to transfer I

identifies the cyclic redundancy check

preamble is a
node that is to bits help receiver verify that
burst of ones
receive the the message was received
and zeros to message without error
enable phase-
lock loops at the start of frame number of bytes end of frame
receiver to lock in bit pattern helps in the data field bit pattern helps
receiver circuits detect to follow receiver circuits detect
start of valid message end of message

Figure 8.10 A data-link frame

MASTER-SLAVE the retransmission is attempted too many times, then the

message is discarded.
As the name suggests one node is designated the master
usually, this is a unique node and also corresponds to the Conclusions
central logic controller of the system. The master controls all
communication activity and the slaves only respond to a The producer-consumer model is very efficient if more
request from the master. This leads to deterministic beha- than one node requires an item of data. However, in many
viour but without any peer-to-peer communications. Much automation applications the source-destination model with
of the fieldbus installed base is PLC systems with a cen- broadcast is adequate. Producer-consumer networks also
tralised architecture and a master-slave protocol. The two need complex set up to allocate data IDs although if tools are
dominant networks in Europe are Profibus-DP followed by provided by the vendor this can be relatively painless.
Interbus S, both firmly entrenched in the PLC-distributed Master-slave protocols only support centralised structures
slave architecture. although slower peer-to-peer connections for nonreal-time
data are possible on some networks. The producer-consumer
TIME-DIVISION MULTIPLEXING capabilities of CAN without using a master arbiter make it
an attractive solution for small networks. Token ring is the
Each node is allocated a time slot on the network when it is most flexible protocol for peer-to-peer communications
permitted to initiate a transaction with a peer node. In the but the deterministic behaviour is not as controllable.
WorldFIP system the time slots are controlled by the bus
arbiter node, which stores a predefined list of data objects to
be produced (note: FIP uses a producer-consumer model).
APPLICATION LAYER
The bus arbiter cycles through the list and broadcasts the
The application layer defines and implements the services that
data identification onto the network - the node that is set up
the network offers each device. The most common services
to produce this data object recognises the ID and then pro-
are read and write. The flexibility of the application layer
duces the value on the network- any number of nodes then
generally incurs large overheads and consequently slower
consume this data. At the time of producing data the node
execution. Indeed, if the dynamic performance is critical then
may request a slot for a noncyclic request. Once all the
many networks bypass this layer for real-time data. Network
cyclic transmissions are complete the arbiter then cycles
data handled in this way is often called cyclic data.
through all the cached noncyclic requests. ControlNet uses
similar media access control.
DEVICE PROFILE
Error Handling
The communication system, layers 1 through 7, manages
In a factory environment, nearby lightning strikes, contact the transfer of data between nodes. The profile or com-
closures, power dips and other events may cause a trans- panion standard is a detailed specification of how this
mitted message to be corrupted. A CRC polynomial inserted data is interpreted or mapped onto device functions. A
at the end of the message is used at the receiving end to common misconception is to assume that, with compatible
determine if the message was corrupted. The data-link layer communications, devices are interoperable: interoperability
detects this and schedules the packet for retransmission. If is only truly achieved if the profile layers are implemented.
196 SIMPLE FIELDBUSSYSTEMS:Modbus

3 SIMPLE FIELDBUS SYSTEMS

There are a number of protocols designed for RS-232 or RS- CONTROL TECHNIQUES' PROTOCOL
485 communications which are still widespread in automa-
tion applications. All Control Techniques' products offer a simple ASCII-based
protocol which allows parameters to be read and written
(Figure 8.3).
MODBUS The response is a single character A C K (hex code 06) or
N A K (hex code 15).
The Modbus serial communications protocol is a de facto
Table 8.1 Query
standard designed to integrate PLCs, computers, terminals,
sensors and actuators. Modbus is a master-slave system Field name RTU (hex) ASCII characters
meaning that one device, the master node, controls all serial
Header none :(colon)
activity by selectively polling the slave devices. Modbus
Slave address 06 06
supports one master device and up to 247 slave devices. Function 03 03
Each device is assigned a unique node address. Starting address hi 00 00
Starting address lo 6B 6B
There are two variants of Modbus: ASCII and RTU. ASCII
No. of registers hi 00 00
mode uses a message format that is printable, messages start
No. of registers lo 03 03
with a colon and end with a carriage return. Error check CRC (2 bytes) LRC (2 characters)
RTU mode uses binary and is therefore not printable. Eight- Trailer none CRLF
bit characters are sent as a continuous burst and the end of Total bytes 8 17
the message is denoted by three and a half character times of
silence. RTU mode messages use half the characters of an Table 8.2 Response
equivalent ASCII message. Field name RTU (hex) ASCII characters
Only the master initiates a transaction. The master is usually Header none :(colon)
a host PC or HMI device since most Modicon PLCs are Slave address 06 06
slaves and cannot initiate a Modbus transaction (the new Function 03 03
Quantum PLCs can act as a Modbus master). Typically, the Byte count 06 06
host master will read or write registers to a slave. In each Data hi 02 02
case, the slave will retum a response message. For a read Data lo 2B 2B
Data hi 00 00
operation, the response will carry the requested data. For a
Data lo 00 00
write operation, the response is used to verify acceptance of
Data hi 00 00
the write command. A special case is the broadcast operation Data lo 63 63
where a write operation can be directed to all slaves. In this Error check CRC (2 bytes) LRC (2 characters)
case, no response message is forthcoming. Trailer none none

The eight-bit address field is the first element of the message Total bytes 11 23
(one byte for RTU, or two characters for ASCII). This field Table 8.3 Query
indicates the address of the destination slave device that
should respond to the message - all slaves receive the mes- Field name Example (hex) ASCII
sage but only the addressed slave will actually act upon it. EOT 04
Slave address 31 1
The function code field tells the addressed slave which
31 1
function to perform. Modbus function codes are specifically
34 4
designed for interacting with a PLC on the Modbus indus- 34 4
trial communications system. STX 02
Parameter 31 1
Two error check bytes are added to the end of each message:
38 8
ASCII mode uses a longitudinal redundancy check (LRC) 32 2
and RTU mode uses a 16-bit CRC check. 33 3
In the example in Tables 8.1 and 8.2, the host PC is initiating Value 33 3
32 2
a read request of three parameters starting with 1.08 from
2E
drive address 06. The starting holding register is 40108 but
37 7
the 4 is dropped in the message string and the rest of the 39 9
register address is entered as one less (0108 becomes 0107, ETX 3
0107 is entered as 006B in hexadecimal). The response Checksum 1 char
repeats the address and function code, but includes the Total bytes 17
values read from the drive.
 Chapter 8.4 197

4 FIELDBUS SYSTEMS

REQUIREMENTS FOR DRIVE APPLICATIONS effects. Some applications will require close coordination of
a number of drives (e.g. each axis in a CNC machine). In
Drive applications dictate specific fieldbus requirements. these applications set points and measured values in all
The following section reviews these requirements. drives must be consumed and sampled synchronously across
the network.
Physical Layer To achieve these dynamic requirements the real-time or
Contactors, power switching devices and switch gear all cyclic data is usually handled differently, bypassing the
contribute to a high level of ambient EM noise. The trans- cumbersome application layer, and presented directly to the
mission media must be immune to this and also be mechani- drive. Moreover the transactions and mapping of the data is
cally robust enough to survive the rigours of installation. generally predetermined during system initialisation to
avoid the overhead of sending addressing information.
The type of media and signalling method will dictate the
limits for the maximum number of nodes on a network and General Message Services
the distance between them (Table 8.4). This should be
checked against the application requirements. Other general nontime-critical access to the network must be
supported. For example, operator interaction, drive set up
Screened twisted pair is generally preferred for cost reasons (PID gains, ramp times etc.), downloading application
and ease of installation. software, general diagnostics and data logging. These mes-
sages are often referred to as noncyclic communication.
Error Detection
The consequence of accepting corrupted data will be cata-
Centralised v e r s u s Distributed
strophic in any application and a reliable error-detection Intelligence
algorithm must be implemented, which will discard cor-
If a communication system with the above dynamic char-
rupted data. Error-recovery schemes are generally too
acteristics is available it may seem desirable to locate all the
complex for small embedded systems and would also require
system control functions in the central controller and simply
redundant data to be transmitted.
transfer demands and feedback cyclically with the drives
Many systems include built-in, high-level error-handling over the network. However, if the drive is capable of
schemes which call for retries ifa frame is corrupted. However, executing some application functions locally then an alter-
for real-time data, which has a limited lifetime, the system is native approach is to use a distributed architecture.
best waiting for the next available sampled data.
In this scenario some control loops are implemented
locally utilising feedback available from the local I/O on the
Dynamic Performance drive. This reduces the load on the network, which is then
only called upon to transfer slower outer-loop references
The bandwidth of a drive varies according to the motor and
which are less sensitive to variations in the deterministic
drive technology used. A servodrive system may have a
response. Although the dynamic requirements of the net-
bandwidth of 1 or 2 kHz. At the other extreme, an open-loop
work are less, flexible peer-to-peer communication is
drive coupled to a large fan or pump may have a system time
needed to allow data to be shared easily between distributed
constant measured in seconds.
processes.
To accommodate the high-performance servo applica-
A distributed architecture has some advantages over a cen-
tions the data will need to be delivered every 1 ms or less.
tralised structure: a powerful controller necessary to cope
Only a small set of real-time data items needs to be trans-
with large levels of computation for the whole system is
ferred at this rate, e.g. speed and torque, control and status.
expensive and the software will invariably be unwieldy and
Moreover, the updates must be periodic and deterministic;
difficult to maintain. Also, the standard I/O which most
in other words the data must arrive at regular guaranteed
drives provide can be utilised very efficiently by a local
time intervals. Ideally the period or cycle of the updates
process without loading the network. Clearly, it may be
should be synchronised with the digital control loops in the
difficult to partition and distribute some systems and a
drives and controller to eliminate any undesirable beating
centralised approach is best (e.g. the interpolation functions
of a multiaxis CNC machine).
Table 8.4 Typical data rates for transmission media

Medium Typical data rate PROFIBUS DP

Screened twisted pair <_5 Mbit/s Profibus DP is a mature fieldbus system that is primarily
Coaxial < 10 Mbit/s used to connect distributed slave devices (I/O, drives, sen-
Optical fibre _< 100 Mbit/s
sors, actuators etc.) to a PLC. Profibus DP is optimised for
198 FIELDBUS
SYSTEMS: Profibus DP

this master-slave architecture and can provide high perfor- capability of CAN. A number of device profiles have been
mance updates when used at the maximum 12 Mbit/s data defined including the A.C./D.C. drives profile.
rate. The dominant physical layer is a multidrop RS-485 bus
using shielded twisted-pair cable. • Assembly objects - cyclic control data is efficiently
packed into the eight-byte CAN data frame without any
Secondary masters (class 2) can be added to the network and protocol information. Assembly objects are usually
can be used for SCADA-type functions only. A master token configured for synchronous exchange with the master
is used to control the master accesses. synchronisation telegram.
There is a wide range of manufacturers in the Profibus user • Explicit messaging - this noncyclic service allows a
group and a number of device profiles (including a drive single object attribute to be read or written. The CAN
profile) have been defined. message carries the protocol information necessary to
identify the object attribute and the value.
Development work on DP to address the synchronisation and
peer-to-peer limitations is ongoing. These enhancements are
both needed for motion control applications. To synchronise CANopen
the slave drives, the master broadcasts a cyclic and equidi-
This is an open-system standard developed by a multivendor
stant clock telegram which all slaves use to synchronise
working group. A number of device profiles have been
internal clocks. The peer-to-peer data is realised using a
defined including the drive speed and position profile.
producer-consumer data model.
• Process data objects (PDOs) - control data is efficiently
packed into the eight-byte CAN data frame without
INTERBUS-S any protocol information- the 11-bit COB ID (CAN
object identifier) is used to identify the data. The COB
Interbus S is a master-slave fieldbus system optimised for IDs must be allocated during system initialisation.
connecting slave devices to a central controller (typically a PDOs can be transmitted cyclically in response to the
PLC). Although the data rate is a modest 500 kbit/s, the synchronisation telegram, or as unsolicited event-based
useful data throughput is high due to the low protocol transmission.
overhead. Slave devices typically provide between four and • Service data objects (SDOs) - discrete messaging to read
ten cyclic transmit and receive data words in conjunction or write an object.
with a noncyclic channel (PCP).
The network uses a ring topology with the master coordi-
nating all network activity. There are two physical-layer CTNET
options: the remote bus variant uses RS-485 twisted pair
with a maximum range of 400m between each node; the Control Techniques developed CTNet as part of the dis-
lower cost local bus uses TTL levels for short spurs. tributed control product strategy. The objectives for CTNet
were low cost, easy to install and set up, peer-to-peer
exchanges to allow data to be shared between drives, and
CAN high performance.
CTNet leverages proven token-ring data-link technology
CAN was originally developed by Intel and Bosch for
with a unique protocol stack having dual cyclic data chan-
automotive applications. Its performance, simplicity and low
nels to provide efficient synchronous transfers for real-time
cost have led to its adoption in many industrial automation
control data. In addition, the general-purpose channel pro-
applications. The main CAN-based standards are DeviceNet
vides occasional or event-driven transactions suitable for
and CANopen.
data logging, diagnostics and set up.
The CAN physical layer is typically very robust but does
A bus topology is used which, with the use of repeaters, can
have a baud rate v e r s u s distance trade off due to the bit-wise
support up to a maximum of 255 nodes. The network uses a
arbitration scheme.
transformer isolated physical layer, which improves noise
immunity and makes installation simpler. The default data
DeviceNet rate is 2.5 Mbit/s.
Originally a Rockwell system, this is now a standard man- When equipped with CTNet, the drive application modules
aged by the ODVA. Most DeviceNet systems operate under can be used to build distributed control systems which offer
a master-slave scheme with a master PLC scanner. How- many performance and cost benefits over an equivalent
ever, this structure does not fully leverage the peer-to-peer PLC-based system.