CANopen

[Veganic 8]

I'm looking at using CANopen where I would normally use CC-link. Can anyone answer the following questions? 1. CC-link continues to run if a slave is removed. How does CANopen react? 2. How CANopen compare to the master/slave set-up of CC-link? At first glance the CANopen system doesn't seem to have a dedicated master. Can each node act as a master/slave?

[kaare_t 155]

You can't really compare CANopen with CC-Link (or Profibus or most other bus'es), but to answer the questions (2. first): The bus always runs in CANopen, and even if a device fails (or even if it short-circuits the bus), the bus is always available and accessible. You can detect if devices are missing by using the NodeGuarding or HeartBeat protocols, but the FX2N-32CAN only supports NodeGuarding.Do you have any other information about the network and/or other devices connected, and what would you expect to send/receive?

[Veganic 8]

Probably about eight nodes. 2 x FX1n sending and receiving 8 words max and 8 bits max each These two parts of the system will not always be connected. 2 x CoDeSys controllers, one of which will be to the main controller in the system. The remaining nodes will be remote I/O. I'm contemplating having one of the FX1n units as a 'subnetwork'. This could have 2 or 3 nodes (the FX1n and CANopen pneumatic valve stations) which could be run as part of the wider network or act alone when disconnected (bus termination could be switched in and out when and if required). CC-link has been great but I always seem to be supplementing it with additional networks to talk to stuff that just doesn't do CC-link. I've considered Profibus and Ethenet. I ruled out ethernet as beyond my capability - I just don't know how I'd get the mitsu PLCs to talk over it. Profibus is ruled out for various other reasons. Any other suggestions gratefully received.

[kaare_t 155]

It sounds like CANopen would be a good alternative, and you could run everything in one dynamic network instead of dividing into several sub networks. The biggest problem with the FX2N-32CAN (and also CANopen in general) is the configuration of both the FX2N-32CAN and/or every other node on the network. You'll need to either configure every node by themselves, or you can use a network manager (NMT Master) which supports some kind of network configuration. Another alternative is to use a software tool (several exists by different vendors) to generate a configuration which you download to each node, or the NMT Master node. Some general notes: I would consider the following before deciding: 1. How many potensial systems? Is this mass-production of a machine? -If not, then I would consider something else. 2. Do you need real-time data updates (and we're talking REAL real-time, not like Profibus or CC-Link which are not the same as real-time)? -If not, then I would consider something else. And by the way; I don't think that an FX1N would do the job if you really need real-time data updates since the scan-time and the FROM-TO bus-time is too slow. 3. If this is a single system (not mass-production), how much money can you spend on learning CAN? -CANopen isn't the easiest system, and you'll have to spend 'pretty much' time just to get the basics of the bus, then you'll have to start configuring and/or programming your units (most dynamic systems require configuration/programming to work - like PLCs, Controllers, ECUs +++). 4. If you're still considering CANopen: Are you sure you can't solve youre problems in any other way (even if this method isn't 'the ideal')? CANopen is a very, VERY flexible system and you can implement almost anything by configuration, but this is also what complicates everything. I'm sorry to be so negative about your idea. I have implemented several CANopen applications, where I have used the FX2N-32CAN as a network manager. I've also used the module to configure other devices on the bus, and/or the complete network system. This is all done by PLC code (depending on the size and connected nodes, the configuration code spans from 1000-15000 steps of code). If you are able to use GX IEC Developer I could send you some of the code I've made and you could take advantage of this. But then again, you would have to get some basic knowlegde of CANopen to be able to use any of the code I could provide, since the code itself is based on functionality and/or terminology of CANopen. If you already know a little about CANopen I've got general code and general function blocks for: -SDO Configuration of specific and/or every node on the network -Network startup and on-the-fly configuration -NMT Commands -NodeGuarding -Custom NodeGuarding -SYNC I've also got some examples of actual (and specific) configuration of (by use of SDO): -FX2N-32CAN -Encoders (Pepperl and Fuchs) -Hydraulic Controller (Danfoss) -ECU (Deif Marine) A good starting point if you would like to learn anything about CANopen is the CiA (nice name.....): www.can-cia.org/ If you have questions feel free to ask me, or continue this thread. Good luck with the decision Edited 17 Apr 2009 by kaare_t

[Veganic 8]

Thanks. There is a lot to take in there. I'm not completely put off using CANopen yet. At the moment I am limited to using something that will talk to FX1n. What would you say to using Devicenet or Profibus instead of CANopen? Edit : I'm tempted to stick to what I have at the moment: A Q plc with integrates the various networks but without having much work of it's own to do. Edited 19 Apr 2009 by Veganic

[kaare_t 155]

Hi! Sorry for the late response, I'm preparing for a 1 month holiday (going to China), so I'm trying to get everything in place at work before I leave..... Well, I'm a fan of the Q-series since I can pretty much do anything I want (except for CANopen... ), but the price is of course a drawback......I would consider the following: Is this a "one-time" application, or is this like a machine with potential of xxx systems? If this is indeed a "small-number" system, I would go for the Q-series because this ensures that you can always get an extra card if any extra needs suddenly pops up. Most often, the extra cost of the Q-series pays off when you start to program since you have the flexibility, and you don't have to spend a lot of time trying to figure out how to solve things or make workarounds in the FX-series.....

[Veganic 8]

I'll PM my e-mail. I like to see the function blocks before I finally commit. Thanks.