Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Tue, 17 Apr 2001 20:36:38 +0200

John Maclean wrote:
> 
> At the APS we are considering adopting CAN for one of our applications
> requiring some remote I/O.
> 
> This particular project needs ~32 binary I/O channels and ~6 low res. (8
> to 12 bit) analog inputs per node.
> 
> Is any one already using a similar off-the-shelf solution, or do people
> generally build their own custom CAN board? Is anyone interfacing
> CANopen or DEVICEnet products to EPICS?
> 
> Someone, (management), also asked me if I knew of anyone who's had a bad
> experience with CAN. I didn't, but as a matter of interest I'd be
> interested to hear of any experiences with CAN, either bad or good.

At BESSY we use CAN for most of the standard I/O, especially all the
Power Supplies.

We developed CAN bus support for EPICS quite a while ago. It implements
a subset of CAL (CAN Application Layer) which I believe is the same as
CANopen. CAL/CANopen includes a connection based protocol with
acknowledged data transfer (but also unacknowledged broadcasts). The
software was designed so that other protocols can be added easily and
multiple protocols can be used at the same time (therefore named it
MultiCAN).

We use VME cards made by esd (VME-CAN2 and VME-CAN4, having 2 and 4
ports resp.) to interface to the CAN bus. On the other side of the cable
there are two kinds of hardware: a general purpose temperature
measurement module (Selectron), and 'self-made' embedded controllers
that connect to the analog and digital (and mixed) I/O cards.

Experiences: Very reliable. You can plug nodes off and on at runtime
without anything getting wobbly. Fast if segment not too long. Warm
reboot supported.

Ben

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System