CAN bus isolation

[alexanderhiam]

Hello all,
I'm working on a board which includes an isolated CAN transceiver. I'm using a DC-DC converter with 1.5kV isolation, and a transceiver IC with 5kV isolation. Because of physical constraints, however, I need the CAN bus to share a 2.54mm pitch terminal block with some non-isolated signals. The distance between adjacent through-holes for the terminal block is about 0.9mm, so (according to Dave's PCB Design Tutorial) that's a maximum of only 150V isolation, or 50V above 3050m ASL.

My question is, is it really worth all the added cost of isolating the CAN bus if in the end I only have 50-150V isolation? What sorts of voltages should I actually expect from a node in a fault?

[Paul Price]

If you have spare pins on the terminal block, just leave these pins unconnected between your two circuits for additional isolation.

[Niklas]

What is the intended application? Do you have any specification that demand X Volts of insulation?

If you don't have any specific insulation voltage specified and just want to break a possible ground loop, then it might be fine with the selected header. One the other hand, if you have a specified insulation voltage, then it might not work with that header at all. Maybe some conformal coating can help, somewhat, but a two header solution with adequate insulation distance would be a better choice.

Box dimensions selected and fixed well ahead of the PCB layout work? Been there, done that :-)

[alexanderhiam]

Thanks guys,

No spare pins I'm afraid, that would be too easy!

It's more or less a development board, so there isn't really any specifications; I was aiming for versatility. I think you're right that the only purpose the isolation would really serve would be to prevent a potential ground loop. It doesn't seem like that's enough benefit to justify the much greater BOM cost. And yes, the box dimensions are fixed, so I can't really add a second header.

Out of curiosity, does anyone know if there are actual recommended isolation levels for, say, an Automotive CAN bus? I'm having trouble finding anything like that.

[lgbeno]

Thanks guys,

No spare pins I'm afraid, that would be too easy!

It's more or less a development board, so there isn't really any specifications; I was aiming for versatility. I think you're right that the only purpose the isolation would really serve would be to prevent a potential ground loop. It doesn't seem like that's enough benefit to justify the much greater BOM cost. And yes, the box dimensions are fixed, so I can't really add a second header.

Out of curiosity, does anyone know if there are actual recommended isolation levels for, say, an Automotive CAN bus? I'm having trouble finding anything like that.

As far as I know there is not such a unanimous standard/recommendation.  In medical equipment there is a general requirement for any part that are in patent or operator contact.  This is iec60601-1. It is really complicated and completely dependent on other working voltages in the system.

I would say CAN bus isolation even if it doesn't meet creepage and clearance, it is worth something from a ground loop prevention but it is expensive...

All in all though if I were you I would skip the cost and sell an additional accessory with proper creepage, clearance and isolation.


Sent from my iPhone using Tapatalk 2

[digsys]

CAN bus Isolation IS worth it in a vehicle situation. I do a LOT of design fault analysis on vehicle CAN systems, including
large trucks from Germany (you'd think they'd get it right). Easiest method is to have the PCB slotted between the
terminals you want to protect. Only need to be past the Pin Pad area. For a vehicle, 80-100V Isolation is fine. You're only
looking for a bad alternator spike, load dump etc in which case other systems will likely fail first. KEEP the isolation.

[alexanderhiam]

Thanks didsys, that's great to know. We'll definitely try to keep the isolation.