e99a036.pdf

MICROPROCESSORS

network (CAN)

intelligent, decentralized data

communications Part 2

The first part

of this article

described the

history, stan-

dardization,

and the basic

setup of the

Controller

Area Network

(CAN) devel-

oped by the

Robert Bosch

Company in

Germany. In

this second

part, the

attention is

focused on the data

transmission proto-

col that determines

the capabilities and

reliability of this auto-

motive digital data

system.

INTRODUCTION

As already stated in Part 1, CAN is a

serial asynchronous communication

protocol that connects sensors and

actuators of electronic control stations

in cars. Among its many functions is a

digital data link. It is an asynchronous

system because each station (also called

‘node’) synchronizes to messages of

other stations on the leading edge of

the first message bit and on subsequent

leading edges throughout the rest of

the message. The ability of any station

to synchronize to another station is

determined by the maximum differ-

ences in oscillator frequencies. Other

factors are, for instance, bit duration,

message duration and composition,

and handshaking.

The most important parts of the net-

work are the physical layer , comprising

the topology of the network and the

link to the bus, and the data link layer ,

which lays down how the data trans-

mission medium is accessed, how a

message is constructed (address, data,

control and protection against errors)

and how the data transmission proto-

col is structured.

By B vom Berg & P Groppe

IN PRACTICE

The exchange of messages between

two network stations may take place in

two basically different ways: station-

Elektor Electronics

10/99

controller area

Arbitration

Control

Data

CRC

Acknowledge

EOF

Intermission

Field

11 Bit

Identifier

DLC

user data, 0 - 8 bytes

15-bit CRC

checksum

ACK

Slot

EOF

7 Bits

3 Bits

3210

Delimits

ACK

Delimiter

990066 - 11

oriented and message-

oriented.

Figure 6. Composition

of a data frame (stan-

dard frame format –

Specification CAN 20A).

to the bus sends a

recessive bit and

another station at the

same time sends a

dominant bit, the dominant bit takes

priority over the recessive bit, that is,

the dominant state is accepted by the

entire bus. The assignment of logic

states to the bus is generally so that a

logic 0 repretransmitteds a dominant

state, and a logic 1, a recessive state.

These arrangements constitute one

of the foundations of the CAN specifi-

cation and will be elaborated on later.

lated. In this, 2 29 =536 870 912 frames

can be handled.

Station-oriented

exchange

In this mode, the sender addresses the

receiver simply by means of the

receiver address, for instance: ‘Station

25 is sending a message to station 37’.

In this way, a virtual link between the

sender and receiver is established via

the bus.

The transmitted packet of data

therefore contains the address of the

receiver station and that of the sending

station. All other stations connected to

the bus ignore the packet since it is not

addressed to them.

The receiving station evaluates the

message and normally acknowledges

its receipt. In case of an error during

the data transmission (no acknowledg-

ment from the receiver), the sender

repeats the message.

RTR (Remote Transmission Request)

bit. This bit, which is always dominant

(0), enables a station to address and

send messages to another specified sta-

tion. This is of great value when certain

data are urgently needed to be

processed (more about this later).

Control field . This 6-bit long section

contains the information as to how a

data frame is composed.

Data packets

The network uses four kinds of data

packets, normally called frames, to

exchange data on the bus: data frame;

remote frame; error frame; and over-

load frame.

IDE (Identifier Extension) bit. This bit

indicates whether a standard-format

frame with an 11-bit identifier (IDE

= dominant = 0), or an extended-for-

mat frame with a 29-bit identifier

(IDE = recessive = 1) is being trans-

mitted.

Data frame

The data frame is used by the stations

to send their data in line with their pro-

gramming. The composition of a typical

data frame, which consists of a single

field, is shown in Figure 6 . This is a

standard frame format according to

Specification CAN 20A. The meaning

of the various terms in the figure is as

follows.

r0 (Reserve bit 0). This dominant bit is

transmitted as a spare bit for future

expansion specifications.

Message-oriented exchange

In this mode, the sender adds to the

message an unambiguous identifier

and sends the message and identifier

via the bus, for instance: ‘Station A is

sending a voltage measurement with

identifier 978’. In this mode, the

addresses of the sender and receiver

are not included.

Such a message is clearly intended

for several receivers connected to the

bus under the motto: ‘Take from the

bus what you need’ (broadcast princi-

ple). The various receive stations must

determine, on the basis of their pro-

gramming, whether the message is rel-

evant to them or not.

DLC (Data Length Code). This 4-bit

long section indicates how many data

bytes are being transmitted succes-

sively in a data field. The CAN Specifi-

cation allows data field lengths of 0–8

bytes, that is, a single data frame may

transmit not more than eight data

bytes.

SOF . This is the start-of-frame bit,

which is always dominant (0). All sta-

tions connected to the bus synchronize

their internal receive stages to the trail-

ing edge of this bit.

Data field . This 8-bit long section con-

tains the data bytes (0–8) to be trans-

mitted.

Arbitration field . This field, which is 12

bits long, contains the data for access-

ing the bus.

CRC field . The 16-bit long CRC field

contains additional information for

protecting the data being transmitted

against interference. For this purpose,

the sender station constructs, accord-

ing to specific rules, a 15-bit CRC

check sum from the preceding data

and sends this, together with the

frames. The receiver station calculates

a similar check sum according to the

same rules and compares this with the

transmitted check sum. If the two

sums are identical (the normal case),

data transmission can commence. If

the sums are not identical, an error

handling procedure is initiated. The

CRC field is limited by a CRC delim-

iter bit which is always transmitted

recessively.

Flow of communications

The flow of communications between

the individual stations connected to

the CAN bus takes place in the form of

a broadcast of event-controlled, priori-

tized (through-numbered) messages or

frames (communication messages).

11-bit identifier . This section contains

the identifier (ID) of the transmitted

frames. The 11 bits allow up to

2 11 =2048 different identifiers to be

constructed, of which only 2032 are

freely available: the remaining 16 are

reserved for certain special functions.

This means that a single controller

area network can process 2032 differ-

ent messages (measurement values,

switch positions, light functions, and

so on). Although this seems a fairly

large number, in many applications it

is not enough. Therefore, an

Extended Frame Format with 29 iden-

tifier bits (CAN 20B) has been formu-

Dominant and recessive bus/bit states

The actual data transmission via the

data transmission medium does not

take place, as usual, in the form of ‘1s’

and ‘0s’, but by dominant and reces-

sive bits. Recessive typifies a bus state

that may be overwritten by a dominant

bus state. So, when a station connected

Elektor Electronics

10/99

Acknowledge field . The 2-bit long

acknowledge field serves to transmis-

sion acknowledgments of correctly

received data frames.

problem. As explained earlier, a reces-

sive bit (1) can be overwritten by a

dominant bit (0).

Figure 7 gives a highly simplified

repretransmittedation of some linking

stages of the bus. Basically, these are

open-collector output stages that are

arranged as wired-AND gates. With

reference to station 1, a recessively

transmitted bit (1) ensures that transis-

tor T 1 remains cut off. This means that

the recessive level is pretransmitted at

the bus. After this bit has been trans-

mitted, station 1 reads the bus status

and recognizes the bit it has transmit-

ted. If then a dominant bit (0) is trans-

mitted, T 1

ing back, station 1 notices that its reces-

sive bit has been overwritten, which

means that it has lost access to the bus

to at least one other station. Station 1

then assumes the receive mode (but

tries to send its message at a later time

again). Stations 2 and 3 continue as

before.

At time j, station 3 sends a recessive

bit that is promptly overwritten by the

dominant level transmitted by station

2. This is noticed by station 3, which

thereupon also assumes the receive

mode (and, like station 1, tries to send

its message at a later time again). Sta-

tion 2 is the ‘victor ’ and can send its

message without fur-

ACK slot . This 1-bit long section is

transmitted as a recessive bit and may,

therefore, be overwritten by a domi-

nant bit transmitted by another station

connected to the bus. It allows receive

stations to send an acknowledgement

of a correctly received data frame. The

acknowledgment bit is dominant and

is transmitted by each and every rele-

vant station upon error-free reception

of messages. Since it is dominant, it

overwrites the recessive bit sent by the

transmitting station. Thus, if the trans-

mitting station receives a dominant bit

during the ACK slot window, instead

of its own transmitted recessive bit, it

‘knows’ that at least one station has

received the message.

The ACK slot window is restricted

by a recessively transmitted ACK

Delimiter bit.

comes on

Figure 7. Highly simpli-

fied repretransmitteda-

tion of how stations are

linked to the CAN bus.

bus line

EOF (End Of Frame) field . This field

consists of seven recessive bits and

serves to terminate the data frame.

Before the next data frame can be

transmitted, the receive stations need

a short intermission to enable them to

process, or at least store, the received

data. The intermission is arranged by a

recessive 3-bit intermission field end-

ing the data frame.

Station 1

Station 2

Station 3

transmitted

received

transmitted

received

transmitted

received

data

990066 - 12

Owing to lack of space, the Extended

Frame Format cannot be discussed; its

principles are, however, the same as

those discussed for the Standard Frame

Format.

and switches the bus line to earth. The

bus line is then dominant (0). Again,

station 1 reads back the bit it transmit-

ted.

Considering the three stages, if one

of them sends a dominant bit, the

busline becomes dominant (0) and the

other stations read this level.

An example will show how the bus

access procedure takes place. Assume

that the stations in Figure 2 are all

ready to transmit their data frames

with three different identifiers:

ther hindrance to the bus.

AVOIDANCE OF

CONFLICT

Since all stations connected to the Con-

troller Area Network bus, two ques-

tions arise:

A closer look at the identifiers

shows that it is the station with the

smallest identifier that gains access to

the bus first: it has the highest send pri-

ority. In other words: the identifier also

automatically contains the message pri-

ority. A message with identifier 0 will

always be the first to be received by the

stations connected to the bus, since it

has the highest priority. A message

with identifier 2032 has a long wait

since it has the lowest priority.

* What happens when several sta-

tions want to send a message at the

same time?

* How is it decided which station can

start and which stations must wait

their turn?

Station 1: identifier 367

Station 2: identifier 232

Station 3: identifier 239.

Clearly, these matters may give rise to

conflicts and to avoid those there is a

special bus access procedure which

must be obeyed by all stations when

they want to send a message. In this,

an important role is played by the

dominant and recessive bits in the

Arbitration Field.

Basically, each sender ‘hears’ its

own transmission to the bus: it sends a

bit, receives it back and compares the

two. If they are identical, transmission

of the message is allowed. If, however,

the two bits are dissimilar, there is a

Remote Request Frame

The remote request frame is an impor-

tant one in the network. Assume that

station D connected to the CAN bus

transmits three temperature measure-

ment data every five minutes with

Identifier 598. This means that the data

field contains three bytes. These mes-

sages are received and processed by

other stations.

However, station G urgently needs

the actual temperature measurement

and cannot in any circumstances wait

for five minutes. It has the facility,

All three start with the arbitration (bus

access) phase by transmitting a SOF bit

(see Figure 8). This is a dominant bit,

and each station reads back its own

(correct) bit from the bus. Then, the

identifiers are transmitted. Up to time

b, all stations send a dominant bit and

all is well. At time c, there is still no

problem. At time d, station 1 sends a

recessive bit, but stations 2 and 3 con-

tinue with a dominant bit. When read-

Elektor Electronics

10/99

therefore, to request the measurements

directly from station D, that is, it can

bypass the data transmission cycle. To

do so, it sends a so-called Remote

Request Frame, which is composed

similar to a Data Frame (Figure 6), but

with some small differences:

DETECTION OF

TRANSMISSION

ERRORS

In a CAN, several means are used

simultaneously to detect errors.

Acknowledgment error detection

In the description of the frame format

(see Figure 6) mention was made of the

ACK slot bit, which is transmitted by a

station as a recessive bit. All stations

that have received the previous frame

correctly overwrite this bit with a dom-

inant bit. The sender detects this and

‘knows’ that at least one station has

received its data correctly.

If the sender detects that its ACK

slot bit is not overwritten, it ‘knows’

that not one station has received its

message correctly. It then shifts its

operation to an error treatment routine

(see later).

Bit error detection

Each and every station receives its own

transmission back. If, therefore, after

the arbitration phase, a station is the

only one that sends a message to the

bus and it receives back a different bus

status than it transmitted, it is clear that

an error has occurred on the bus. The

station then shifts its operation to an

error treatment routine (see later).

• The identifier of the station to whom

the request is transmitted (here,

598) is entered in the identifier

field.

• In the DLC field, the number of

useful bytes contained in the

requested message (here, 3) is

entered.

• The Remote Transmission Request

(RTR) which is dominant (0) in the

Data Frame is made and transmit-

ted recessively (1). This is a typical

identification of a station that

requests data direct from another

specific station.

• There is no data field in the Remote

Request Frame: the DLC field is fol-

lowed immediately by the CRC

field. In other words, the Remote

Request Frame is composed like a

Data Frame but with 0 bytes of data.

Stuffbit error detection

The CAN specification states clearly

that when in a data frame more than

five bits of the same value are trans-

mitted in sequence (for

instance, seven times a

0 in a field), each and

Figure 8. Diagram-

matic repretransmitte-

dation of how a station

does access the bus

(Arbitration).

Format error detection

In this, use is made of the fact that the

network format has several fields that

must always have a fixed content: the

CRC delimiter, the

Acknowledgment

Delimiter, and the EOF

The transmitted Remote Request

Frame functions as follows. All stations

connected to the bus receive the frame

and recognize by the set RTR bits that

a station has requested specific data

from another station. Station D recog-

nizes that the identifier in the Remote

Request Frame is the same as its own

identifier and immediately sends its

response in the form of a Data Frame

with the requested data.

bit instant

bcde f gh i j k l

Identifier

Station 1

000101101111

Station 2

000011101000001...

Station 3

000011101111

bus state

000011101000001...

Station 2 wins and keeps transmitting ...

ERROR DETECTION

AND REMEDIES

One of the most striking properties of

the Controller Area Network concept

is its uncanny capability of detecting a

multitude of errors during the data

transmission and react to them accord-

ingly. It has a Hamming Distance (also

called signal distance) of 6. The signal

distance between two binary words of

the same length is the number of the

corresponding bit positions in which

the two words have different bit val-

ues. For instance, the signal distance

between 11011010 and 10000110 is four,

since the 3rd, 4th, 5th, and 7th bits

(counting from the left) are different.

In a CAN data are transmitted per-

manently at a transmission rate of

500 kbit/s. Every 0.7 s a one-bit error is

caused by external interference. The

network operates eight hours a day,

365 days a year. The built-in protection

against errors in a CAN guarantees

that in 1000 years of operation only

one error will not be detected. Errors

can and do, of course, occur, but once

they are known, they can be remedied.

Only unknown errors can cause false

measurements to be processed.

log. '0'

dominant bit (bus state)

log. '1'

recessive bit (bus state)

990066 - 13

every group of five bits is followed by

a complementary bit (here, a 1, of

course). This introduced bit, which, of

course, contains no information what-

ever, is called a stuffbit. At the receive

end, these bits are removed from the

data stream, so that only the original

message is processed.

The stuffbits may readily be used

for error checking. If the receiver

detects more than five sequential bits

of the same value in a frame (but not in

the EOF field), it is clear that this can-

not be right and that an error in the

data transmission has occurred which

has inverted one or more bits. The

receiver then shifts its operation to an

error treatment routine (see later).

field are always composed of recessive

bits. If a dominant bit is detected, this

can have been caused only by a data

transmission error. Here also, operation

is shifted to an error treatment routine

(see later).

ERROR TREATMENT

The error treatment routine in

response to a data transmission error

takes two forms.

In the first place, frames in which

an error has been detected are imme-

diately rejected by the relevant station

and not processed. Secondly, if any sta-

tion within the system detects an error,

it immediately transmits an error frame

that consists of six dominant bits (=

error flag) and an error delimiter of

eight recessive bits. The result of this is

that all recessive bits on the bus are

overwritten, so that six dominant bits

remain. This is, however, a contraven-

tion of the stuffbit rule that not more

than five sequential bits may have the

same value.

CRC error detection

This consists, as already described, of

an evaluation of the CRC check sum at

the receiver. When the received and

calculated check sums are dissimilar,

the receiver shifts its operation to an

error treatment routine (see later).

Elektor Electronics

10/99

All other stations connected to the

bus detect this error condition and

treat the frame just transmitted as

faulty, reject it and also send out an

error frame. In other words, a station

that detects an error purposely muti-

lates the entire transmitted frame so

that all other stations connected to

the bus receive a faulty frame. This

means that an error local to a station

is immediately communicated to all

other stations. The motto of the net-

work is that all stations receive cor-

rect data that can be processed as

required, or all stations receive faulty

data that are rejected. The original

sender detects, of course, that the

frame it transmitted is mutilated,

adjust its message and resends it

after a short while.

Table 3.

CAN 20A

CAN 20B

Maximum number of identifiers

2 11

2 29

Number of stations (nodes)

Data transfer rate

5–125 kbit/s 5–1000 kbit/s

Number of permissible bytes per frame

0–8

Maximum length of a frame

117 bits

13 bits

Maximum bus expansion

see text

Table 3. Comparison of CAN 20A (standard frame

format) and CAN 20B (extended frame format).

SUMMARY

The specifications of the two CAN ver-

sions, CAN 20A (standard frame for-

mat) and CAN 20B (extended frame

format) are compared in Table 3.

Although the Controller Area Net-

work is a powerful and highly reliable

system for data communications, the

reader and prospective user may well

ask how it can be turned into a practical

application. There are dominant and

recessive bits, an 11-bit identifier, a

15-bit CRC check sum, a 1-bit delimiter,

a 7-bit EOF field, a 6-bit error frame,

and many more. None of this resem-

bles the 8-bit or 16-bit data structure of

the microcontroller.

So how is it possible to program

according to the network protocol?

Here, the future constructor need not

worry. There is a plethora of ready-

made, inexpensive building blocks

available for the network. It is this sup-

port by IC manufacturers for the CAN

that has made the network so popular

in such a short time.

ERROR INSIDE A

STATION

What happens when a station itself

becomes defect, is damaged, oper-

ates with an inaccurate transmission

rate, or is the only station that gets

interference? Such a station would

permanently send out error frame

and so disable the entire network.

The CAN concept has adequate pro-

tection against such an occurrence,

but space prohibits describing this in

this article.

The next instalment will deal with

these building blocks, with the pro-

gramming according to the CAN pro-

tocol and with practical application of

the network.

[990066]

C ONSTRUCTION GUIDELINES

Elektor Electronics (Publishing) does not provide parts and compo-

nents other than PCB s, fornt panel foils and software on diskette or IC

(not necessarily for all projects). Components are usually available

form a number of retailers – see the adverts in the magazine.

all the components in the correct position? Has correct polarity been

observed? Have the powerlines been reversed? Are all solder joints

sound? Have any wire bridges been forgotten?

If voltage levels have been given on the circuit diagram, do those

measured on the board match them – note that deviations up to ±10%

from the specified values are acceptable.

Large and small values of components are indicated by means of one

of the following prefixes :

Possible corrections to published projects are published from time to

time in this magazine. Also, the readers letters column often contains

useful comments/additions to the published projects.

E (exa) = 10 18

a (atto) = 10 –18

P (peta) = 10 15

f (femto) = 10 –15

T (tera) = 10 12

p (pico) = 10 –12

The value of a resistor is indicated by a colour code as follows.

G (giga) = 10 9

n (nano) = 10 –9

M (mega) = 10 6

µ (micro) = 10 –6

k (kilo) = 10 3

m (milli) = 10 –3

h (hecto) = 10 2

c (centi) = 10 –2

da (deca) = 10 1

d (deci) = 10 –1

In some circuit diagrams, to avoid confusion, but contrary to IEC and

BS recommandations, the value of components is given by substitut-

ing the relevant prefix for the decimal point. For example,

3k9 = 3.9 kΩ

4µ7 = 4.7 µF

color

1st digit

2nd digit

mult. factor

tolerance

Unless otherwise indicated, the tolerance of resistors is ±5% and their

rating is 1 ⁄ 3 – 1 ⁄ 2 watt. The working voltage of capacitors is ≥

black

–

50 V.

brown

10 1

red

×10 2

±2%

orange

10 3

–

In populating a PCB , always start with the smallest passive compo-

nents, that is, wire bridges, resistors and small capacitors; and then IC

sockets, relays, electrolytic and other large capacitors, and connectors.

Vulnerable semiconductors and ICS should be done last.

yellow

×10 4

–

green

10 5

0,5%

blue

10 6

–

violet

–

grey

–

Soldering. Use a 15–30 W soldering iron with a fine tip and tin with

a resin core (60/40) Insert the terminals of components in the board,

bend them slightly, cut them short, and solder: wait 1–2 seconds for

the tin to flow smoothly and remove the iron. Do not overheat, par-

ticularly when soldering ICS and semiconductors. Unsoldering is best

done with a suction iron or special unsoldering braid.

white

–

gold

–

10 –1

silver

–

10 –2

10%

none

–

±20%

Examples:

brown-red-brown-gold = 120

Ω,

Faultfinding. If the circuit does not work, carefully compare the pop-

ulated board with the published component layout and parts list. Are

yellow-violet-orange-gold = 47 k

Ω,

Elektor Electronics

10/99

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: