Ethernet IP
July 2004
Amazing changes have taken place in the automation community in the last few years.

With the PLC-5 soon going into 'Silver Series' (an estimated timeline of 6more years), I have been wondering what is going to replace the Remote I/O and DH+ protocols that were so common in PLC-5 systems. Would it be DeviceNet, ControlNet, or Ethernet?

Ethernet had been out of the question because of it's lack of 'determinism'. Random retries during collisions was not a good way to run I/O, although the speed of Ethernet is extremely fast compared to other networks (100 Mbps maxtypical).

DeviceNet is an Open Master/Slave type network. Many different vendors support DeviceNet products, which made this network look promising. This network has disadvantages, however: Many customers have complained about the software being unstable, power cycles are sometimes required to get thenetwork to run if it quits, and the maximum speed is only 500 Kbps and only supports 64 nodes per network.

ControlNet was both deterministic, and fast. It has a speed of 5 Mbps, supports up to 100 nodes per network. With ControlNet, a packet (in someway) can be 'guaranteed' to get from the I/O to the processor at specified intervals (ie every 5ms or 20ms) depending on how much bandwidth was available on the network. The ControlNet network itself could replace both the DH+ and the Remote I/O Network. The ControlNet didn't seem widely accepted yet, although it looked like the network of the future....

But then something happened...Tests were done with Ethernet to determine if network collisions are really a problem. Engineers began to ask themselves, "If Ethernet runs 20x faster than DeviceNet, would a few network collisions even be something to worry about?". "If we have an isolated network, just for I/O, we can run a protocol that everyone already knows, with very inexpensive hardware.", they said.These thoughts brought about the idea of Ethernet/IP (industrial protocol). Since the I/O network is isolated, a special protocol can be used that makes this network deterministic eliminating the previous disadvantages of Ethernet as long as the I/O network is isolated from the Enterprise network. Over the past few months, I have had the priveledge of working with Ethernet/IP products. I have found them very easy to configure (Maybe easier than Remote I/O was), and very stable. I would pick Ethernet/IP over DeviceNet or Remote I/O any day! Imagine your remote chassis just having an Ethernet card (with an IP address) plugged into a hub. The processor connects to this chassis (using the IP address or host name) in much the same way that Remote I/O Worked. Messaging between processors can be done using a producer consumer model! (No confusing message statements in ladder logic). It looks like this maybe the way of the future! If you have any questions, please feel free to email me! I'm glad your newsletter system is back up and running!!!