CAN Bus protection (100VDC)

If you are making the CAN signals 'ride' on the power, then they will have a **common mode voltage** of whatever that power is, and you'll be hard pressed to find an IC with 311V of common mode range (the peak voltage of 220Vac) and can withstand a similar negative excursion. It's not just protection, it's data recovery in the presence of such high common mode voltages that will be the problem.

CAN has a nominal common mode voltage of 2.5V relative to circuit common (CAN High has excursions above this, CAN Low goes below this by about a volt or less). You might get a floating circuit, or perhaps even capacitively couple the signal and then shape it back up again, but I would see this as a design effort in amplifiers, not simply plopping down ICs.

Cheers

PeteS

I think Stan was talking about the power being on separate wires in the same cable as the data lines and was wondering about protection against the power lines shorting to the data lines.

I think he's going to need a fuse (maybe a polyswitch) and watch out for the effects of the introduced capacitance and resistance. Raychem has some telcom protection devices that might be worth looking at.

Robert

It has occured to you that this might have been a breathtakingly bad idea, right?

It's also quite probably the only feasible way out of that pitfall. You asked for problems --- and, well, you got them.

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

Un bel giorno StanV digitò:

Instead of just isolating the CAN bus, you could do it for all the electronics connected to it, by using isolated AC-DC power supplies. It's very unsafe though: with a short between HV and CAN, either the electronics and the CAN bus would be at HV potential, and if someone touches it - after all it is just an harmful CAN bus... - he probably won't be isolated from the ground. :)

--
emboliaschizoide.splinder.com

What kind of fuses do you have feeding the 100 V and 230 V lines ?

If we are talking of only a few amperes, connect one reverse biased rectifier diode from Gnd to CAN_H and an other from Can_H to Vcc rated for at least a few times the fuse rating. Do the same for CAN_L. Put a

5.5 V 30 W zener between Vcc and Gnd. Assuming the logic Gnd is connected to the 100 Vdc and 230 Vac ground/neutral, this will blow a 1-5 A fuse in the 100 or 230 V feed.

This will remove the offending voltage from the system. Unfortunately, the large rectifier diodes have a quite large capacitance, but putting

2-3 in series will reduce it, but of course the clamp voltage will increase.

If the high power circuit has larger fuses, then you would have to put a fuse (maybe 100 mA) in series with CAN_H and CAN_L on _each_ node and use smaller rectifier diodes rated about 1 A to Vcc and Gnd, thus, the stray capacitances would be lower. A 5.5 V 1 W zener between Vcc and Gnd should be sufficient.

This circuit will blow the series fuses when the CAN nodes, but it does not disconnect the high power supply, so you may be zapped by dangerous voltages, when you try to figure out, why there is no traffic on the CAN bus :-).

Paul

..but you can consider change to opto-isioation ?

If these are on the same cable and share 'coal face' connectors, then yes - hmmmm...

You have a couple of protection issues. Highest priority should be personnel : if mains _does_ get onto the CAN, what fuses where ? [can 220VAC end up on the PCBs and CAN network, with a broken GND trace ]

Fusible resistors could be one solution, if you can find one that promises to keep back mains.

Transformer isolated busses are another solution, as are isolated BUS drivers.

-jg

Cost or no cost, I'm not sure opto will even save you. CAN being a

2-wire differential bus, odds are your HT will short to just one of the CAN lines, not both. So now you have up to (almost) 400V *between* the two CAN lines. Opto isolators will contain the damage, but you've definitely lost your CAN transceivers.