Does running multiple CAN buses in the same cat5 cable sound like a bad idea?

[e100]

Currently I have a CAN bus running at 20k baud that collects data from multiple sensors. Would adding adding a second CAN bus to talk to another bunch of sensors at a faster rate (using another 3 wires) be likely to cause problems?.  The two busses would have their own ground return wires, and they would only be connected together at the power supply.

[capt bullshot]

should work

[T3sl4co1l]

Ground return?  Presumably, on the other remaining pairs?

Just mind that you'll get a little noise / impedance between those.  That is, between the different pairs in the cable.  Nothing CAN can't handle, by a long shot, if it's wired up in the usual way.

The pairs themselves should be quite good: the twist rates are different just for this sort of reason.  Other CAN buses will be jealous of the red carpet you're putting it on here.

Tim

[schmitt trigger]

Theory says it should work but I would still perform a simple experiment:

Run data thru one of the CAN lines, while monitoring the other in recessive state.

With a dual channel scope you should trigger from the active channel. That way, any noise that is synchronized to it you'll know is the crosstalk.

[max_torque]

As long as you run each bus through the correct pair of cables (ie the ones twisted together) then i'm sure oyu won't have an issue, especially down at just 20kbaud.

CAN gets bundled up, un shielded in much worse conditions than those existing in a Cat5 cable with CAN running on the other cables!

I guess you can't just up the baud rate, because that's the easiest option to get a higher throughput?

[linux-works]

20k baud does not sound very fast (cars often are at 500kb).

it should work.  I'd run a test to see if any checksum errors are seen or if you get any NAKs.  try various traffic patterns using a traffic generator to see if crosstalk becomes a real issue.

but 20k is quite slow, so that's not very hard to do.

for fun, I'd try various combos of can-h and can-L on the pairs of wires; perhaps running some 'out of phase' (so to speak) may be beneficial.  testing will show.