RE: Long, Low speed CAN bus

> From: John Dammeyer
> ....
> A quick and easy way is with a pulse generator and scope.
> https://www.edn.com/electronics-blogs/scope-junction-
> blogs/4431974/Quick---dirty-cable-length---impedance-measurement
> ....
> So take your cable.  Start with a 5V 80uS pulse and look for the =
reflection to
> determine the impedance and termination needed.  And look at the edge
> rounding.  Next transmit 6.25kHz (80uS ON, 80uS OFF) and look at the =
signal
> at the other end with respect to rounding of the edges.  Maybe use a =
CAN
> driver to do this instead.

Interesting, For a long time I have wanted to build a "Time Domain =
Reflectometer" (TDR) for diagnosing cable issues.  My plan is to drive =
the cable with a MCP2551 driven by a 555. =20

BTW: My two largest sites have gone to 1k termination resistors on 6 bus =
segments ranging up to about 2,000 feet and they are working perfectly =
(no bus errors reported by the system).

> What you will find is that if you don't have the cable properly =
terminated the
> reflection may or may not impact your messages.  Generally the =
reflection is
> only an issue when it disrupts messages at specific points along the =
cable.
> ....
> What's worse is it's not just a pulse but a sequence and all but one =
message
> may be received perfectly but one will be intermittent. Very hard to =
find the
> reason.

I realize that difficulties with reflections can be 'iffy' and depend on =
exact node location, etc. so I will be monitoring things.

I can't explain why at the moment (more work with the scope may help =
find out) but, for whatever reason, when using very low data rate over =
long distance it seems that amplitude is more important than reflections =
- but again - more monitoring/testing needed ...

> ....
> Another thing often overlooked is the CAN DC ground.   For those kinds =
of
> distances and cable resistance you have to address common mode
> difference between the signal (relative to ground) and the signal at =
the other
> end relative to its own ground.

Awhile ago we encountered this.  There were 'ground' differences between =
the sites (each has a ground rod) that I think were caused by current =
flowing in the common 'ground' wire connecting the sites.  The sites are =
connected with a pair of #16 THHN for power as well as the CAT-5.  The =
entire system draws ~1.5A @ 28vdc.

Things got a lot better when we addressed this.

> ....
> So for those kinds of distances using isolated CAN drivers where the =
power
> supply for the drivers accompanies the signal may work better.  Ie. =
Opto
> isolators on the TTL side.

Yes, I have looked at these for two reasons: a) Lightning protection, =
and b) 'ground' reference issues.

I used CAN as opposed to RS-485 because it gives me one interrupt per =
message as opposed to one per character with message handling and bus =
arbitration for free.  The issue I have run into is that it is more =
difficult for customers to deal with when it doesn't work because the =
issue may be at any of the controller locations along a long length of =
cable.

My nodes report Bus-Off rates, etc. and they 'suspend' the driver for 10 =
seconds if the node detects that it is trying to send unsuccessfully =
(bad receiver causes MCP2515 to continuously resend the message, =
swamping other bus activity) but then the customer has to diagnose the =
issue.

Most of the failures are in cable connections or the lightning/surge =
protection between the cable and the MCP2551 or ground reference issues. =
=20

I have considered adding a high-impedance ADC to CH & CL so that the =
node could detect issues such as CH/CL short to ground, CH/CL reversed, =
ground reference issues, etc. in terms that the customer can relate to.
=20
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]>