Remote IO connections over fiber ring

Hello,

I inherited a project where the previous guy designed a comms ring to go around the site to connect all the existing PLCs. He specified an awful lot of redundancy to be added because his original plan was to leave all PLCs in the same room and install IO all around the site on this fibre ring.

The fibre ring would eventually have VLAN's to separate Safety, SCADA, and PLCs. The site settled on Rockwell and it would be a much bigger cost now to move on to another vendor so we're sticking with them, meaning that the network switches were also specified as Stratix (57 or 5400).

When reading forums, etc... yesterday I came to an interesting post where the IO running on top of a ring like this would suffer if the ring broke since the reconfiguration time is higher than what the IO modules would like to see.

Personally, I dislike the approach very much as I like to see physical segregation and Device Level Rings wherever possible, but with all the hardware out on site, I'm tempted to take a chance on some systems, but not if something like this has the potential to shut a plant down.

Has anyone here done something like this? And what is your opinion on the best way to do it? Mind that we're moving from SLCs and a bunch of Micro and Compact Logix to a few Control Logix in the near future.

Back in 2004 I had a costumer that wanted a control system that would never fail, now "Never" is a big word on a Slaughter-line, but he was not that concerned about the price which where nice

The costumer had a system with 4 SLC's communicating up to a joint database, so the PLC's the network and the DB all had to be in perfekt order for the system to work, this gave way to much downtime and loss of production data.

I merged the 4 SLC's into one Compactlogix PLC, added an internal buffer in the system that would hold production data for 24 hours, this was back in the good old DeviceNet days, so All remote IO was setup with DeviceNet.
When I was done I doubled the whole system: UPS, PLC, DeviceNet and DB so I had two identical PLC's getting the same data in, On PLC 1 I had an output going to an input on PLC 2, if this was high it would block any outputs from PLC 2

At first we thought this might be a bit overkill, but it has earned itself back many times over, the beauty of the system is that every time we had to make changes I could make them in PLC 2 and we where all happy with the result, I would copy them into PLC 1
From all the test we have made in this system, I have not found anything that is more redundant, i don't loose any scantime when switching over, rock solid

What ring protocol are they currently using? That would mostly determine the network effects caused by a network failure. It sounds like you'd need a redundant ring, where all data goes both directions on the fiber to prevent reconfiguration delays.


I've used HSR for this in the past.
https://en.wikipedia.org/wiki/High-availability_Seamless_Redundancy

I am following the Cisco/AB/Panduit document for 'Deploying a Resilient Converged Plantwide Ethernet Architecture'. You can find the document on the AB site just search for "enet-td010_-en-p.pdf" or the above title.


I tried to upload it here but its to large.


Its a very well thought out design, although the Resiliency can make it pricey, but you can always scale it down a little.



Like your talking about I have small locations like MCC rooms that have one or two DLR for the VFDs and other IO in the MCC. Just be sure to use shielded cable in those.

mk42 said:

What ring protocol are they currently using? That would mostly determine the network effects caused by a network failure. It sounds like you'd need a redundant ring, where all data goes both directions on the fiber to prevent reconfiguration delays.


I've used HSR for this in the past.
https://en.wikipedia.org/wiki/High-availability_Seamless_Redundancy

Click to expand...

Not sure yet as the scope definition didn't went that far from us and I'm yet to power up the first switch. Granted, none of the IO is to be done just yet so I do have time to ponder a bit.

JaxGTO said:

I am following the Cisco/AB/Panduit document for 'Deploying a Resilient Converged Plantwide Ethernet Architecture'. You can find the document on the AB site just search for "enet-td010_-en-p.pdf" or the above title.

Like your talking about I have small locations like MCC rooms that have one or two DLR for the VFDs and other IO in the MCC. Just be sure to use shielded cable in those.

Click to expand...

Cheers, I think I have it and my understanding is that they don't go for this type of comms with IO. There's also the bonus that on a DLR you don't have to worry where the network cable is to be plugged which is a problem when using VLAN's.

Many jobs are being specified here in Oz with the redundant Schneider M580 PLCs and Ethernet IP I/O - the Schneider Ethernet cards can be set up in a ring and when the ring is broken at 1 point there is no change in scan time or anything else. If the remote module dies you lose I/O at that point.
I looked at another method using an Omron dual redundant PLC and Omron Ethernet I/O with redundant switches on 2 networks set up in 2 rings - each switch would connect via an Ethernet cable to the Omron Ethernet IP I/O module - there are 2 ports - and would also communicate with each other and if a switch failed the other network ring would take over.
the grumble was that the dual redundant CPUs and power supplies were on the same rack so I had to go with the Schneider solution.
Schneider are out there selling there system very hard - you do not want to know the cost LOL!!!
They were fibre rings with a wired CAT 6 cable out of the wire/fibre optic converters - they were in my opinion a weak link!
Omron for there dual redundant system have dual direct fibre optic networks available and they automatically change over = seamless.