RE: Guaranteed Valid Messages

Chuck,
Dig deep into the CAN specification and you will find that the CRC used for
CAN messages is robust but does not detect all errors.  So there is a small
likelihood that a message with more than two error bits could create a
message that would pass the CRC and be received as a correctly formed
message.  It's unlikely you are having this issue.

The next issue is more complicated. Remember how CAN works.  Any single node
receiving a message can overwrite the message with an error flag of 6
dominant bits if it detects an error.  Again look closely at the CAN spec
and there is a spot where the error could be missed which is why TOGGLE
messages are not a really good idea.

So again if the message is somehow improperly formed or damaged someone will
catch it and everyone will see that error flag.

Finally we're into the actual CAN device here.  How might it fail?  The
state machine inside is pretty straight forward.  The bus has to be idle for
a specific number of bits.  Any noise on the bus that doesn't meet the bit
definition criteria restarts the bit counter for bus idle.

Once the falling edge of the start bit is detected and it's still dominant
at the sample point the state machines launch into their bit assembly and
destuffing algorithm.    This is where a FORM or STUFF error can be detected
and later at the end of the message a CRC Error.

As I understand it the interval after the ACK bit is when the CAN device
checks to see if the message contained in the MAB matches the filters.  If
there was an error flag the message is always discarded.  You'd never get a
message that was damaged with an error flag since that doesn't tell you
anything about the message other than now 6 or more dominant bits in a row.

Perhaps what Microchip means is that if the Filters are set with values but
the Rx bits are set to 11 then all messages are received irrespective of the
filter and mask values.  When set to 00 then both 11 and 29 bit IDs are
received based on the filter and mask values.  So your setting of 0b11 isn't
really incorrect.  Unless Microchip is passing something else through.

Best thing to try would be to set the bits to 0b00 and set your masks to
allow all messages.  Ie. Filter bits are don't care.  If you still see the
same symptoms then you have a hardware problem.

John Dammeyer



From: [email protected]
[mailto:[email protected]] On Behalf Of Chuck
Hackett
Sent: May-04-15 7:20 PM
To: [email protected]
Subject: [CANLIST] Guaranteed Valid Messages

Just to be sure my understanding is correct .

Assuming a CAN bus controller is correctly configured (e.g.: set to receive
ONLY valid messages as in my previous post), there is no way that an invalid
(i.e.: corrupt/damaged) message will be received . lost maybe, but NEVER
corrupt . regardless of bus disturbances, excessive length, too many loads,
etc.

If an invalid message is received it must have been invalid when it was
given to the sending CAN controller.

True?

Cheers,

Chuck Hackett
"Sometimes it is the people that no one imagines anything of who do the
things that no one can imagine ."
7.5" gauge Union Pacific Northern (4-8-4) 844
http://www.whitetrout.net/Chuck



--
Archives and useful links: http://groups.yahoo.com/group/CANbus
Subscribe and unsubscribe at www.vector.com/canlist/
Report any problems to <[email protected]>