Why not?
Even if there was a ground connection between two CAN interfaces, this would only connect both *isolated* grounds together. I have here a 400A welding transformer which is controlled over a CAN bus by a panel PC. The CAN connection does have a ground, but also in this case, it only connects the isolated "grounds", which are just the V- from the isolated DC/DC converter. So no real ground, just an equal reference to both ends.
Meindert
Some reputable manufacturers make isolated converters without even a ground terminal on the RS-422/485 side, for instance
Some of their older products had a separate connection for the shield to the isolated signal ground, but the newer versions do not have that external connection.
Paul
(floating receivers and transmitters without GND connection)
well, that's a rather special device likely not built with standard transceivers (e.g. they have to detect idle state safely). They should know and use the necessary balancing technique.
BTW there is a discrepancy, in the text they state EN
61000-6-3 but in the DoC EN 61000-6-4 .Oliver
-- Oliver Betz, Munich despammed.com might be broken, use Reply-To:
(floating receivers and transmitters without GND connection)
That's what you want - a "reference" terminal to carry the current caused by common mode noise over the isolation barrier.
Oliver
-- Oliver Betz, Munich despammed.com might be broken, use Reply-To:
Scope the signals for shape, logic-analyze them for timing. For a rough test I would probably just stress the bus a little extra to see how far from breaking point I was.
Speaking of bus length, that would mean to introduce a bit of artificial signal delay on top of the existing system to see how much headroom there's left. I.e. splice some extra tens of meters of cable length into the bus, and see when communication starts to fail.
Oh, and maybe add some more of your type of CAN nodes (listening only) to the bus to see when their electrical load overburdens the bus.
Shift the sampling point around to see how much tolerance you have there.
Oh yes, expanding the bus until it breaks sounds like a good test. We can use a can analyzer and use that to see at which length the error rate starts to rise significantly.
Then how to scope the signals. We don't have a differential probe, so we could just measure between H/L with a single probe on the (floating) bus. Or measure both with respect to ground and use the scope math, or get a differential probe. Or use an INAMP ...
-- Stef (remove caps, dashes and .invalid from e-mail address to reply by mail) While your friend holds you affectionately by both your hands you are safe, for you can watch both of his. -- Ambrose Bierce, "The Devil's Dictionary"
Adding nodes and cables is a good idea, but use only one node in listen mode, the others should be in normal (send/receive) mode.
When a node in normal node detects an illegal frame,e.g. bit stuffing error or CRC error, it should start sending the error frame, which will also destroy the bit pattern for other nodes, thus avoiding the situation that some nodes close to the transmitter would get the frame correctly while others nodes at a large distance would not get the (command) frame.
One idea could be inspect the only node in listen mode. If the controller has some error frame counter or status line, it could be used. Alternatively look at the receiver signal between the CAN transceiver and CAN control. You can look at this (digital) signal with an ordinary probe and visually detect the error frames.
Paul
Or you could use two identical standard probes, and use the sum function on a pair of scope channels.
The usual thing on differential high speed lines is to observe the "eye" pattern. Eye open, good. Eye closed, not so good...
Regards,
Chris
This will not work - you must not probe the bus with a low impedance (the GND connection) even if it's floating.
this works.
Sapphire Si9001 and clones are cheap (approx. 300EUR) differential probes, they have a somewhat limited high frequency CMRR but might be suitable.
This might be better than the differential probe. The AD8130 looks very interesting because it's unique design results in a very good high frequency CMRR - 70dB@10MHz! Respect the input CM and diff range!
Oliver
-- Oliver Betz, Muenchen (oliverbetz.de)
Thanks all for your replies. The system is still being built and we won't have access to the required hardware for a while (I started the thread with cable selection ;-) ). Measurements will be performed when we do get some hardware.
-- Stef (remove caps, dashes and .invalid from e-mail address to reply by mail) The solution to a problem changes the nature of the problem. -- Peer
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.