>> Yes but isn't the CAN version used in cars the kind with one wire and
>> shared ground ? I am under this (strong) impression. And so are most
>> interface chips.
>
>I don't know.  That's possible since every penny counts and they can
>afford a slow bit rate, but I would have expected differential due to high
>noise environment.  All the CAN driver chips I've seen are differential,
>and this is what is described in the standard as I recall.  However, that
>is merely one implementation of CAN in any case.

Looking at the CAN driver schematics I can see why Peter thinks they may be
a single ended system. It would appear that they are not a "true
differential" drive with one side going to both positive and negative
voltages. The driver chip has a transistor which pulls the CANH line to
positive, and another to pull the CANL line to ground, but it looks like the
"other" state is open circuit. It would seem the receiver is a true
differential item with the threshold set high enough to minimise receiving
any noise.

....

>> The kind of CAN that is most commonly used is the one wire plus ground
>> kind. Twisted or not. And it does not like transients. The open
>> drain/bit squashing part that does the arbitration requires that the
>> bus be relatively high impedance at least part of the time.
>
>I haven't seen that type of CAN system.  All this proves is that it's
>possible to create bad implementations of CAN (I'm not even sure this
>would be true CAN anyway).  I was objecting to your statement that CAN
>inherently has these problems.  CAN is designed to work with differential
>signals, and this is widely supported by various semiconductor
>manufacturers, including Microchip.  I'm quite sure Phillips has a can
>driver chip similar to (probably even drop in equivalent) to the MCP2551.
>I think Microchip is merely a follower here.  If I remember right, they
>claim superiority in power consumption, but are otherwise trying to be as
>compatible as possible with industry standard CAN drivers.

I cannot see how it could be run as a one wire plus ground, and be called a
CAN bus (see my email from yesterday). I have just had a look at the
Microchip MCP2551 and the Philips PCA82C250 and PCA82C251 transceiver chips,
and they are indeed pin compatible. The MCP2551 is claimed to be suitable
for 12 and 24V systems, Philips only claim the PCA82C251 to be suitable for
24V. Have not done a close comparison of the specs otherwise, but it would
seem they are drop in replacements.

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics