On CAN cable impedance and RJ45 connectors

Hi all, I recently stumbled upon this project and I love the work you’re doing here. I just performed my first read-through of the UAVCAN Standard V1.0-alpha, so I might not fully grasp all the details (yet). Anyway, one thing I noticed is that there’s no mention of the required cable impedance in the Physical Layer chapter. There is a nice reference to the ISO11898-2 standard, which does specify the required impedance (as far as I know). My question is whether you’d think it would be a good idea to include some statement about this in the UAVCAN Standard.

The reason for my question is that the referenced CiA 103 is quite specific about this:

The characteristic impedance of the cable shall be 120 Ω; ±10 % in the frequency range from DC up to 500 kHz.

A fact that may complicate matters is that for short cable-runs, the characteristic impedance of the cable is not as important (in my experience). So users may be tempted to ignore this requirement since it “already works”. Also, controlled-impedance cables are generally quite expensive compared to a simple twisted-pair. Also, I’m not sure about the availability of controlled-impedance cables with “Micro-connectors”.

A common solution (I’ve seen) for budget CANBus systems is the use of RJ45 connectors combined with standard Cat5 Ethernet cables. It turns out that these cables already have a spec on characteristic impedance, but 100 Ohm instead of 120 Ohm. In this case, a pair of 100 Ohm termination resistors could be used. As you know, these types of cables/connectors are very cheap and widely available.

Some general sources about ISO11898-2 which contain some insight into the design-considerations:

Hi Gijs! Thanks for joining.

You are correct to point out that the physical layer specification is somewhat lax.

I believe that @scottdixon had his hardware engineers look over the spec and yield some constructive criticism which may also include the impedance considerations, we did not have the time yet to process the feedback.

We are still in the early days so it is probably optimal to postpone the introduction of hard constraints into the Specification until we have received more substantial empirical feedback from the industry.

Alright, sure! I wasn’t sure what part which part of the specification was still WIP :slight_smile: . Is there another place where you gather feedback like this, or is a forum-post like this the best way?

Anyway, I think any “hard constraints” are not even required. The CiA 103 spec only briefly mentions this 120 Ohm cable-impedance once, so I think this might be a good example. Users are of course still free to ignore these specs if they feel it’s not required in their application.

Also, I think table 7.5 is only valid if this cable impedance is correct, so a quick mention can’t hurt.

You are right. I just created an issue to track this: https://github.com/UAVCAN/specification/issues/81

This forum is the right place for this.

I do think the hardware part of the specification could be better. I have some pending feedback but, unfortunately, I’m not able to provide all of the feedback I’d like to due to confidentiality constraints so having another hardware engineer provide feedback would be really helpful.

I’d love to help where I can :slight_smile:. Just let me know how/where you need my input.