Re: Implementing CANopen => Help

From: Paul Keinanen (keinanen_at_sci.fi)
Date: 09/30/04 

Date: Thu, 30 Sep 2004 11:31:25 +0300

On 29 Sep 2004 10:32:03 -0700, gerard0sweeney@gmail.com (Gerard
Sweeney) wrote:

>I'm currently working on implementing the CANopen protocol from the
>ground up on a CAN-enabled processor.

Are you building a CanOpen master of just a slave ? If you are making
a slave, do you need to pass the interoperability test or make a
device that works in some specific systems ?

For a simple slave, look at the "predefined connection set" and if it
is sufficient, a lot of the OD stuff can be hardwired. At a minimum,
you will have to handle the NMT (start/stop), PDO:s and answer some
expedited (single frame) SDO request with a constant single frame
reply. You need the ability to set the node number on the device,
since it affects the CAN identifier numbering.
  
>However, I'm having some trouble
>understanding how (and where) to actually start.
>I have the Application Layer and Communication Profile (CiA DS 301).
>I've been reading about the object dictionary quite a bit, but am
>unsure what exactly the OD does. How is it used? What do I need to do
>to actually implement the OD?

You are gong to need the OD, if you intend to implement a CanOpen
master with freely configured nodes (slaves). Usually you start by
reading the EDS (Electronic Data ***) files for each node type or
the DCF file for each individual node and build some tree structure in
memory to hold this data.

Paul

Quantcast