RE: MCP2515 - Recovering from Error-Passive state
"Chuck Hackett" <[email protected]>
| Newsgroups | gmane.comp.hardware.bus.can |
|---|---|
| Message-ID | <[email protected]> |
> From: John Dammeyer
>
> ....
> You haven't talked much about your protocol. If you still have the
potential
> of two nodes sending the same ID with different data you are hooped. You
> will not solve the periodic bus disturbances. And they will happen often.
I am in the process of eliminating this problem ...
> ....
> It's been my experience that with larger networks the nodes tend to drift
> together when sending messages so you end up with blocks of nodes
> arbitrating. If you aren't filtering properly then often nodes end up
with
> buffer overruns when you think you have a bus that is only loaded at 10%.
> What you really have then is a bus that is 100% loaded for 10% of the
time.
I noticed this too ('bunching' of bus activity). I didn't consider it a
problem under the assumption that the CAN protocol would work it out but, as
you say, it can result in the firmware having to deal with 'bursts' of
receive messages.
My nodes do not use the 'filter' option because there are not enough filters
available. Any given node may require a large number of different message
IDs depending on the configuration a user selects.
> If you have some sort of heartbeat message sent by each node on a periodic
> basis then have them include their TEC REC counters within the heartbeat.
> Collect that information over a period of time and you might see a cluster
of
> nodes with problems.
In the process of adding that now ...
I also have the nodes periodically reporting who they are 'hearing' from and
with what error rate to help diagnose bus problems.
In one location currently the nodes near one end (but separated by maybe 300
feet) are having trouble talking to each other but other nodes on the same
bus (about 1,500 feet long) seem to be happy. If it was only the fact that
the bus was too long for the data rate (I am looking into this) I would
expect that the nodes close to the other end would also have trouble (if the
problem is 'late' ACKs destroying dominant bits, etc.).
Cheers,
Chuck Hackett
"Good judgment comes from experience, experience comes from bad judgment"
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]>