Centralized or Distributed IO

I'm just curious about what you all would do.

I have a project that is in the preliminary design phase. It consists of replacing a Modicon 984 with a ControlLogix PLC system. The M984 has four racks all in one main cabinet with a fifth rack in a cabinet right next to the main. The machine occuipies 2100 sq ft (200m^2).

The field wiring is sloppy, not to code, and is going to be replaced.

I am leaning towards using distributed IO (CNET and DNET) with a single CLX rack at the main panel. Abouht half of the IO is on the same side of the machine as the main panel, but it is concentrated in an area about 15-25 feet (5-8m)from the panel (about 40'[12m] as the conduits run). About 20% is at the operator console on the opposite side of the machine 45' (14m) away. 10-15% of the IO is in an MCC about 10' (3m) from the main panel. The rest of the IO is scattered about the machine.

What are your thoughts and ideas on distributed -vs- centralized IO?

It depends on the labor cost issue & response time. Can the remote I/O respond in a timely manner as required by the process, and how much labor savings are YOU going to realize by not having to add terminals & cables etc... I usually need to see mabey 20+ remote I/0 to start thinking about going with a remote base.

Going with the previous post, how well you know CNET and DNET communications. Yes, it does reduce wiring which reduces labor but if you have a huge learning curve then it might be better to go direct. When you finally get CNET and DNET to communicate then your home free but it is a little painful to get to that point. You may be able to use CNET for everything and DNET for the MCC.


Hope this helps....

I've used both CNET and DNET before, though with not quite so many nodes, so the learning curve is not an issue.

Its pretty much a given that it will be DNET to the MCC, it has C-H Advantage starters, so its likely that we will just add a DNET adaptor to each of them and be done with it.

We've been using DNet for about 6 or 7 years now and have had great success. Since there doesn't seem to be any learning curve for you I'd say go for it. Obviously, if your system requires extremely fast response times for things like Web registration and/or servo positioning I'd hard wire those functions. Reduction of conduit sizes and the associated wiring can easily pay for the additional costs associated with Distributed I/O, if you shop around! Plus, you can't beat the flexibility of a Distributed I/O scheme in an ever changing production environment. Having said all that, I'b be sure to be careful if you were looking at upwards of 64 nodes and a bunch of explicit messages in the mix. In that case you might think about having two seperate networks to ease the load. As for CNet, I'm affraid I haven't used that yet.

Alaric said:

I'm just curious about what you all would do.

I have a project that is in the preliminary design phase. It consists of replacing a Modicon 984 with a ControlLogix PLC system.

Click to expand...

ControlLogix for Modicon 984 huh, I'd go for Quantum.

Depending on your application requirements have your considered using plain old ethernet between chassis? No extra networking software required. Simply give each Ethernet card its own IP address and use an industrial switch to join the ENBT together. With the 100MB/s speed and the switch to avoid collisions performance should be super fast.


I have two applications at my work where we are using Ethernet between chassis and have yet to have a problem.

Some personal preference involved...

I have always preferred centralized I/O with a lot of wire, myself.

But that probably comes from many years of troubleshooting relay panels by myself before people heard of PLC's.

A wireless laptop would be an aid to me, specially of DNET.

I won't admit it quite yet, but probably the way to go.

regards.....casey

I would definately use distributed. I pushed for it at my company. We began to use PLC's for communications between the control houses and the operator consoles on drill rig power controls we build. We wired the house to a cabinet instead of the plug panel. This replaced 5 or 6-20 conductor cables and several intermediate plugs. The concern was the noise from the 1800 HP dc drives of which there would be 4 or 5 of them. We now use remote I/O in each drive cabinet. The MCC still hard wires to the PLC cabinet which is the end cabinet of the MCC. We use the dual profibus network to the operator console due to "rough necks" having to handle the cables during rig down and rig up.

On one design, DNET is used from the main control house to the auxiliary house.