> CAN is both an electrical specification and a bus protocol. It requires
> special silicon to support it fully (i.e. you have to buy someone's
> chips
> - perhaps Microchip's).

No special silicon is required by the standard.  You could implement a
complient CAN controller from discrete parts if you wanted to.  Off the
shelf silicon has emerged because CAN is a useful concept and the
automotive industry (which invented CAN) has large purchasing power.

I don't see "buying someone's chip" as a big deal.  CAN is widely
supported by many manufacturers, including Microchip.  Certainly this is
no worse than having to buy a PIC from Microchip.  You can get both in one
package in the 18Fxx8.

> The CAN bus is not very well protected against transients.

Pure bullshit!  CAN is specified as differential, and most implementations
are.  See NMEA 2000 (layered on the CAN standard) as an example.  It is
intended to run 100s of meters in a ship using unshielded twisted pair.

> This means that it will eventually blow up in my experience,
> taking out one or more devices.

More bullshit.  Any bus that goes reasonable distances is subjet to common
mode transients and ground offsets.  Proper design in these evironments
dictates some sort of isolation.  The CAN standard leaves this up to
individual implementations, but it in no way prohibits you from doing so.
Again NMEA 2000 is an example of CAN where devices are encouraged to be
opto-isolated from the bus.


*****************************************************************
Embed Inc, embedded system specialists in Littleton Massachusetts
(978) 742-9014, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu