THE MAGAZINE FOR COMPUTER APPLICATIONS
Circuit Cellar Online offers articles illustrating creative solutions
and unique applications through complete projects, practical
tutorials, and useful design techniques.

RAMBLINı MAN

by Tom Cantrell

Start ı Little Network ı Mini-Message ı LO BAT ı Caravan ı Sources and PDF

MINI-MESSAGE

Letıs look at the overall way a LIN setup works. As shown in Figure 3, the basic unit of activity is a message frame. The master initiates a message transfer by issuing a header that consists of three parts.

Figure 3ıThe master issues the header, and the appropriate responder (i.e., the master or a particular slave) gives back the data in due course. (enlarge)

First, the master holds the bus low for at least 13-bit times to get all the slave nodesı attention with the SYNCH BREAK. At the end of the SYNCH BREAK, the master sends the character 0x55, which (with alternating ones and zeros) acts as a SYNCH FIELD. Each node times the bit transitions in the SYNCH FIELD to adjust their own data rate. With everything in sync, the master sends the IDENT field, which includes a 4-bit identifier, a 2-bit designator that specifies the amount of data to transfer (1, 2, or 4 bytes), and two parity bits. The parity scheme seems like a bit of a head-scratcher but, besides simply checking the integrity of the IDENT field, also ensures that patterns 0x00 and 0xFF never appear on the bus.

At this point, the master has requested the appearance of a particular data item defined by the 4-bit identifier and length (1-, 2-, or 4-byte) designator. Now, a responder puts the data on the bus, protected by a simple checksum rather than a complicated CRC.

Typically, a slave node provides the response. However, itıs important to understand that the response isnıt to the master, but to the bus. In fact, the master may supply the data in response to its own header, which allows it to send its own data on the bus. Furthermore, everything on the bus, the master and all slaves, has the opportunity to look at the response. For example, a door lock solenoid and window motor could both go into action upon detecting a door key lock/unlock message.

How does LIN deal with the possibility of collision? It doesnıt. The protocol dictates only a single responder. Each node is required to monitor its own output in order to detect and log the occurrence of a collision, but thatıs it. Thereıs no collision recovery (i.e., arbitration) mechanism built-in because that would compromise timing predictability. Of course, the master can request collision and other error-status reports (parity, checksum, or no response) from time to time if desired. Otherwise, the strategy is basically to try again next time (i.e., bad messages are just thrown away), which is arguably the best course of action when dealing with lights, switches, and such.

Notice that the spec doesnıt limit the time allowed between the header and data or between each data byte. Instead, it simply dictates a maximum length for the entire message frame. Thatıs nice because it means even a slow micro can handle the protocol, yet the overall message timing remains bounded.

PREVIOUS • NEXT

Circuit Cellar provides up-to-date information for engineers. Visit www.circuitcellar.com for more information and additional articles.
For subscription information, call (860) 875-2199, subscribe@circuitcellar.com or subscribe online. ıCircuit Cellar, the Magazine for Computer Applications. Posted with permission.