Fiber through Hubs

[Dryno]

I'm working on a project for a client to upgrade existing air compressors by individual 'slave' controllers on each compressor and one load-shedding master controller. These controllers communicate via a vendor-proprietary protocol over fiber. Since fiber is involved, the client's IT department is involved. They propose to route the fiber via existing hubs to the compressor controls. Those hubs serve the plant's existing network(s). We'll need ~7 miles of fiber if we go through the hubs. I had assumed that we'd run the fiber direct from controller to controller which would require ~1.5 miles of fiber. Again, this is what I would call a stand alone system, so I don't see why the fiber should be routed through hubs. What benefit would this provide? Is this 'normal'networking practice?

 

[itsmoked]

No but brain dead IT departments are normal practice...

You need to talk to the engineer in charge of the compressors and discuss the pros and cons for this particular site.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[ScottyUK]

Put in a fibre switch local to the controller. Run fibre to the controller from the compressors. This would allow each compressor to communicate with each other and whatever the supervisory controller is. No need to go to the central hub.

Maybe it's just the IT dept trying to get a long fibre installed to a remote location on your budget?

Add a few extra cores to the fibre over and above what you expect to need: we don't use anything smaller than 8-core these days.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[stevenal]

Perhaps the network provides redundancy, so that a fiber cut does not result in loss of control.

 

[Dryno]

Thanks guys,

I'm also concerned with the hubs; if they're true hubs and not switches, what about collisions and delay? I've asked the compressor controls vendor (CCC) for some particulars on the communications protocol because the sales guy kept telling us that it was 'high speed'.

Also, the fiber will be a redundant install so if one is damaged or otherwise compromised, we'll have a backup. The fiber is ~90% existing, I just found out.

 

[ScottyUK]

If you are going for redundancy I suggest a industrial ring switch from someone like Westermo or Hirschmann will give the most resilience while tolerating an industrial environment. Unless you are moving an outrageous amount of data - which seems very unlikely in the application - then a 100MB/s ethernet ring will be close to realtime in performance. Westermo's R200 looks a good starting place:

http://www.westermo.com

If you are moving truly prodigious amounts of data then you could probably find a 1GB/s equivalent but consider that a typical modern DCS architecture is based on 100MB/s redundant ethernet and reliably handles tens of thousands of data points.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[itsmoked]

Yes hubs blow.. They should be switches.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[robertjo24]

You stated in the original post that “These controllers communicate via a vendor-proprietary protocol over fiber”.
My question would be, are the controllers connected to one another via Ethernet over fiber, or are they actually fiber ports on the controller that are direct connected together?

If it is the latter, I would have to assume that utilizing the fiber hubs would not work.

I guess you will have to have the vendor confirm, but it sounds like you would need an isolated set of fibers for your controls.

I know this is how many of the Siemens and Allen Bradley controllers and PLCs work.

 

[itsmoked]

Also it dawns on me that if it's some proprietary protocol a switch/router may choke on the compressor messages whereas a 'dumb' hub wouldn't know the difference.

Keith Cress
kcress -

http://www.flaminsystems.com