ControlLogix Redundancy on Ethernet/IP network

We're just looking at setting up a redundant ControlLogix chassis pair. The pair communicate to a single I/O rack via Ethernet/IP, and we will also have a SCADA system talking to the redundant chassis.

All the examples I've seen for redundant pair show 2x EN2T cards in the controller chassis. One card for the connection to the I/O rack. A separate card for connection to the SCADA/HMI.

Is this really necessary? or just 'best practice'?

I would've thought you could have just one EN2T card in each controller chassis. Then connect everything (SCADA, I/O rack, controller chassis) to a managed switch sitting in the middle. SCADA and I/O comms can be separated via VLAN.
Would this work, or have I missed something?

answer - yes, it'll probably work fine - It sounds like you already know why it shouldnt be done that way.

When I see configured switches on an automation network - I get scared. What will happen if that switch gets rebooted, reset or replaced ? Will you have a good backup of that configuration ? How much do those things cost again ?

If the IO is on its own physical network, you can put any cheapo switch and get the traffic flowing in the case of failure. If its a managed switch it needs to be configured before installation - that is another engineer-level task that doesnt need to be a part of component replacement imho.

that being said - I'm sitting right next to multiple racks of configured switches as part of a DCS system. I also have a 24hr hotline to vendor tech support, some IOS experience myself, and the cell phones/emails of the 'good' service techs that can help me with cisco issues.


side question - why redundant processors without redundant IO ?

-John

I thought that IO in a control logix redundant system had to use control net? Has this changed?

Thanks for your reply Johnster.
You can also say, if you're forking out the cash for redundant controllers then a couple of extra EN2T cards are really neither here nor there. Your points regarding managed switches are valid too.

Good question re. why redundant processors without redundant IO ... with the usual answer of 'that's what the client specified and they're not moving on it'.

Put me in the camp of more distribution is better...We typically have three ENBT's....One for local I/O, one for Drives (usually talking to 30 to 40 drives), and one for Plant Comms (SCADA, PLC-PLC Messaging). No managed switches, just cheapie cisco dummies.

BTW, your twisted pair media is not the bottleneck for comm performance. It's the ENxT to Backplane communications. Another reason for multiple ENxT's even if they are on the same subnet and physical media network.

Oceansoul said:

I thought that IO in a control logix redundant system had to use control net? Has this changed?

Click to expand...

It all boils down to the question of what the redundancy is provided for... Is it to keep the plant/process/machinery in operation, or to retain the process data.

Single channel I/O is common in many process industries, where the retention of the batch process data is most critical. ControlNet "redundancy" modules (CNBR) introduce a single point of failure. The redundancy they provide is seamless switching of cable systems.

A truly 100% "redundant" system would have redundant I/O, parallel valves, pumps, actuators, etc. Not many systems go that far.

Oceansoul said:

I thought that IO in a control logix redundant system had to use control net? Has this changed?

Click to expand...

AB now has enhanced redundancy with Ethernet/IP. Much faster failover time and overall better performance.

http://literature.rockwellautomation.com/idc/groups/literature/documents/um/1756-um535_-en-p.pdf