|
Issue
98, September 1998
Networking
with DeviceNet
by
Jim Brady
Start
• Sorting Them Out
• New
Breed •
Motivation •
Can •
Message Reliability •
DeviceNet Connections •
Device Net Messages •
Stringing Messages Together •
Some Real Messages •
Object Library •
Conformance Testing •
DeviceNet Standards •
References,Sources,PDF
Message
Reliability
The
CAN controller calculates a 15-bit CRC value for the received
data and compares it against the CRC it received. If an
error is detected, the node originating the message is
notified so it can resend the message.
If
the originating node sends too many messages for which
it gets an error back, it goes offline. That way, a bad
device won’t crash the entire network.
Your
CPU detects this event by reading the CAN chip’s
status register. You have the option of staying offline
or initiating an error-recovery sequence.
If
the CAN controller originating a message doesn’t
hear back from at least one other device that the message
was correctly received, it will resend. Thus, a lonely
node will just sit there and send over and over.
When
you do get a message from your CAN controller, you know
it’s correct. And when you tell the CAN controller
to send a message, it keeps trying until the message gets
through. Pretty good, considering all this is handled
by one $8 chip.
To
make CAN into a usable network, you need a way to string
messages together, establish connections, and handle errors.
That’s where DeviceNet comes in.
|
