CAN design principles + Earthing (double question)

Hi there, just have a couple of questions to make sure I'm doing the right things. I've been working on the pinout documents for the looms for my racecar and the last part of this is just making sure that my CAN and earthing layouts are correct before I draw everything properly.

The way I understand CAN is that there should be a main trunk then each branch should be kept to a minimum length. Therefore, I've done a layout as per below. This does mean that in certain sections of the harness that the bus doubles back on itself to avoid having a branch that is too long (the isolator, for instance, is half way down the vehicle). Does this look right? Most diagrams show the ECU at the end of the trunk but in terms of layout, it's actually better if I put the PDM on the end. I assume this won't be an issue.

Also CAN B is just a little one between my dash and the steering ECU. If this is quite short, do I still need a terminating resistor at both ends?

Secondly, after rewatching the earthing section I think I'm pretty happy with the earthing layout, but below is a simplified diagram of how I'm doing it.

I have the Battery Negative pins of the ECU going to a star point on the intake manifold along with the earths from the coils etc. There is currently a large earth strap between the engine and the body and the battery is earthed at the rear of the car to the chassis. The battery negative pins on the PDM I'm planning on just taking to the chassis locally near the PDM, along with any other device powered by the PDM except for the ECU and coils, of course. There are obviously more devices than drawn, this is massively simplified. Sensor grounds of course will use their own gnd directly back to the sensor ground pins on the ECU. Does this look ok also?

Thanks!

yes, you will need a terminating resistor in even a very short CAN bus.

Both ends though David or just one?

If it is less than 500mm in length you can sometimes get away with a single resistor, but you may get reflections of the signals coming back from the end of the bus that doesn't have the resistor on it. Some CAN devices will cope with this without issues, others may have problems. I would add the resistors at both ends to ensure that you have done everything to reduce the chances of noise induced errors.

Perfect, thanks Stephen - makes sense. Is there any issue having an extra resistor half way down the trunk? For example the dash, which has a resistor as part of the factory terminated loom. I can obviously remove this, but just wondering.

We have done test looms with 12 resistors (one resistor per device) and not seen any issues with the data, but it isn't recommended. One extra is better than one missing.

Great info, thanks for the reply.

Those MoTeC CAN devices are strong! I had a Honda Racing ECU, and it would get CAN errors whenever a UTC was added to the bus. That might have been the 3rd or 4th resistor (because the harness was designed with 2 resistors, then we found out some devices had fixed resistors installed.) I fixed it by replacing the ECU with a MoTeC M130.