Rockwell 1756-EN2TR redundant cable questions.

I have a spec that requires cable redundancy whether or not it makes sense. Also, I have to have redundant Processor Chassis where only one is powered at a time. I already have the Redundant processor system running on another project and that is easy. The Chassis are connected in a star network using 1756-ENTR cards and the two processors have the same IP address and as only one can possibly be on at a time, it all behaves itself perfectly.


My question is twofold. First and easiest:
Can I set up a 1756-EN2TR where both ports from every device runs to an unmanaged switch in a star network structure and provides cable redundancy (unplug any one cable and still have everything run)? (I know that this still would shut everything down if any one actual device failed except the PLC processor which can be manually switched to, and it primarily would be cable redundant.)


My second question might possibly require a diagram, but I'll try in words:
Can I set up a Device Level ring where I use two unmanaged switches, one for each ring direction coming out of the 1756-EN2TR on the PLC processor chassis so that I can switch between chassis and have one completely turned off at any point?

I would personally prefer to use the first method and I would just buy everything and test it, but Rockwell makes their equipment out of pure gold and if it is impossible I might have to do something stupid, like use devicenet.


Thank you,


Glen Osborne

Great questions !

The first existing system you describe is not "redundant" in the usual sense of the word, and certainly it's not ControlLogix Redundancy. I think the best term would be "cold standby".

The "Both ports of a 1756-EN2TR to an unmanaged switch" network will not work correctly. Both ports of that module have the same IP address but unique MAC IDs, so you will promptly get a duplicate IP fault but not a spanning tree fault or network loop storm.

But hang on to that idea; I'll revisit "Parallel Link" cable redundancy.

The "Each port of a 1756-EN2TR to an unmanaged switch" network will not work, but it's a little harder to explain why. There won't be an IP conflict, because the two ports won't ever see one another's traffic, but each ControlLogix will have two physical paths to the same adapter/IO device and it won't be able to figure out which one to use.

Thinking more about your specification, I see the challenge:

If you built a DLR network as it's intended to be designed and operated, the powered-down ControlLogix would mean that there is always a broken link which means the other devices in the Ring don't have media fault tolerance.

And if you used a 1783-ETAP so that a powered-down CPU wouldn't break the ring, then the CPU doesn't have media fault tolerance because it only connects to the 1783-ETAP with one cable.

Which brings us around to "Parallel Redundancy Protocol", a fairly new variation of the 1756 EtherNet/IP bridge modules.

There is a 1756-EN2TP module available. Instead of implementing a Device-Level Ring, it implements Parallel Link Redundancy, where two Ethernet cables connect the 1756 module to separate switches (similar to your second description), and if one of them fails the other takes over, quickly enough to support the fast automation I/O, safety, and motion protocols used by ControlLogix.

I'm not as familiar with PRP as I am with DLR. I'm not sure if you can do it with an unmanaged switch; I would have to research it. I think that ordinary I/O might work with an unmanaged switch and PRP, but things that depend on precision time would not (motion and safety).

Here's the technical application note for PRP with ControlLogix:

https://literature.rockwellautomation.com/idc/groups/literature/documents/at/enet-at006_-en-p.pdf

Ken,


Thank you, you've definitely saved me from some major frustration.


I'll look into the PRP option. I suppose that there is some chance that I would need 3 Stratix 5400 to get that to work with a PanelViewPlus in the mix. Either that or Rockwell is trying real hard to sell me lots of those expensive switches as every diagram they have is chock full of Stratix 5400 links.



https://imgur.com/gkfdkPo
Here is the diagram for my possible fix using a device level ring. Though perhaps you've already shot that one down.



https://i.imgur.com/eZ7j1zG.jpg
Perhaps I can do it with one Stratix 5400 managed switch and the 1756-EN2R in the same way that this diagram uses the 1756-EN4TR pairs.Does that seem reasonable?



Thank you for your help,


Glen Osborne

Perhaps I can do it with one Stratix 5400 managed switch and the 1756-EN2R in the same way that this diagram uses the 1756-EN4TR pairs.

Click to expand...

Nope, EN2TR does not have Redundant Adapter functionality. Must have EN4TR

Is the "cold standby" method and the desire for un-managed switches driven a hard-specification, a budget, or a goal of configuration simplicity ?

Is this a specification to which you're still bidding, or a project that has been awarded ? I've had clients who wrote an impossible technical requirement into a spec to weed out two kinds of weak bidders: ones who lacked the technical expertise to recognize the trap, and ones who would say "yes" to things they couldn't provide.

If I were building a system to meet this network and functional specification, I would propose building a DLR network with 1783-ETAP switches at the ControlLogix location, and accept that the controllers themselves are subject to a single cable failure, but that cable is well-protected inside a control cabinet and has a lower probability of failure than the CPU, network module, or a switch. The cables that actually leave the cabinet are the ones that participate in a single-fault-tolerant DLR loop. There's no question that it would work as desired because it's the simplest DLR design.

You might even save a few bucks by proposing 1756-L8xE controllers with a built-in single Ethernet port, instead of a 1756-EN2TR.

The system that you show in your first post will not work correctly with an unmanaged switch. You could use a non-DLR switch, but you would need a managed switch that allowed you to disable MAC table learning on the related ports, and set up the proper Quality of Service settings described by the ODVA's DLR application whitepaper.

This is a job I'm doing already, and am running out of time to do since they took a full month to reply to my submittals.



That cold standby method is standard on a number of movable bridges around Florida and is the method used on the bridge now and the contract wiring diagram.



That is the current method used on the bridge right now, the bulk of the job is just to upgrade from PLC-5 installed 20 years ago to 5000 without changing too much of the operation.


I've already completed one job on an identical bridge, replacing the existing cold standby in kind with a system that uses a regular star network and I've seen it on another half dozen bridges at least. They switch from A to B processor every week. It works fine.


This is from the contract plans:



To understand why I'm trying to finagle this weird double-wire star network, I submitted the same star network that I did on the previous job, but since this time they drew in a note on the system that says:"Control system utilizes one of the two PLC processors and a redundant control network, control backup is provided by by a second PLC processor and a complete (existing to remain) hardwired relay control system for manual, maintenance operation but with complete safety interlocks."

Which, in conjunction with their wiring diagram caused the person who approves the submittals to suggest using the 1756-EN2TR which I was afraid wouldn't actually work and you've pretty much confirmed that at least it won't work with the dual wire star and an unmanged switch. means I could order everything now before waiting for another submittal if I was sure it would work. Otherwise I might have to wait another month I don't have for them to come back if I submit something completely new.


I might see if I'm allowed to talk to their engineer (not always allowed in this bureaucracy) and propose getting rid of the "cold standby" and just using an actual redundant processor system and a device level ring.



(Sorry if these paragraphs were jumbled but I'm simultaneously working on 4 jobs and putting out fires all around heh.)

Thanks very much for the context. I've done a little work on moving bridges and live within rowing distance from one of the world's busiest bascule bridges, and I sympathize with the challenges of engineering for government clients. If things go right I'll spend next spring at Merritt Island.

If it were me, I would try to submit a cold-standby (consistent with their other systems and practices) system with 1783-ETAP to connect to 1756-EN2T (or 1756-EN2TR if they have them, just using one port).

That will give them a single-fault tolerant network; any one cable or device in the ring will allow the rest of the system to keep running. You would need to lose the 1783-ETAP *and* the active PLC (or its connection to the ETAP) to have a loss of control.

The bridge between Merritt Island(Oops, Cocoa) and Cape Canaveral that goes over to where they load the cruise ships was the second full electrical refurbishment of a moving bridge I worked on from start to finish. The Journeyman I was helping at the time and myself did pretty much the whole install with a few people brought in from other jobs for wire pulls etc. Good times. heh.