ControlNET between two redundant controllers

[elmatador]

Hi Guys,

I have run into a little snag here will laying out my system architecture on paper. I have two redundant sets of controllers that need to talk to each other over controlnet. There are no other nodes on this link...which is a PROBLEM. You need a minimum of 2 nodes per redundant chassis pair.

Citing: http://literature.rockwellautomation.com/idc/groups/literature/documents/um/1756-um523_-en-p.pdf

Page 35

This is what I have:

PLC1-PRI -> PLC1-SEC -> PLC2-PRI -> PLC2-SEC

What do u put in between for controlnet nodes? Does Rockwell have some sort of Dummy Node to handle this type of situation?

 

[elmatador]

bump

 

[Ken Roach]

Is this ControlNet network a "supervisory" network, separate from the I/O networks run by these two Redundant controller pairs ?

I build most of my supervisory networks on EtherNet/IP, but you can do it with ControlNet just as well.

If you install these nodes as described, during switchover of one Redundant pair, the network will stay working. If both Redundant pairs switch over at exactly the same time, the network will undergo a "bump" as the 1756-CNB or -CN2 modules go into "lonely" mode then reconnect. That "bump" will break any Produced/Consumed Tag connections and might require retransmission of a few unscheduled messages, but the network should reconnect in just a few seconds.

 

[Ken Roach]

To answer your last question specifically: No, RA doesn't have a "dummy node" for this function, but I have used a 1794-ACN(R) adapter to provide the required ControlNet Moderator function in a similar system.

The basic idea is that a ControlNet network can stay "live" as long as there's more than one device connected. A single device goes into "lonely" mode, which is what you want to avoid.

When the 1756-CNB or 1756-CN2 modules in a Redundant Controller pair switch nodes, they both electrically disconnect from the network for a few milliseconds, because they need to re-initialize the ControlNet transceiver.

If there was just one other node on that network, it would go into "lonely" mode in those milliseconds, and the network would have to go through a Moderator election and re-synchronization process. It takes only a few seconds, but that's plenty of time for an I/O connection to fail.

This is why you can't have a Redundant ControlLogix with just one rack of I/O.