Overcoming Ethernet’s limitations
Ethernet was designed as a way for computers to communicate over a single, shared cable so that businesses could connect systems to support programs and transfer data. The beauty of Ethernet is its ability to transfer copious amounts of data over the network very quickly, and to provide information to remote users.
As industrial manufacturing markets began to see the advantages that Ethernet provided, many companies began to explore the possibility of implementing Ethernet on the manufacturing side of their operations. However, standard TCP/IP Ethernet was neither designed nor intended to be a control level network. Thus the debate began over how to make Ethernet work in the industrial control world.
Networks using the original Ethernet protocol standard communicate information by sending and receiving data packets via the Ethernet frame. A data packet can contain from 64 to 1,055 bytes of data at speeds of up to 1 GB per second and beyond. However, manufacturing industry users may have difficulty with this arrangement due to how the data are sent.
Ethernet is designed to send large amounts of data less frequently than control networks. Also, Ethernet is not deterministic %%MDASSML%% meaning that when data are sent from one point to another, a user cannot determine when the data will be received by the intended location. In most commercial applications this is not an issue because %%MDASSML%% unlike in manufacturing %%MDASSML%% the data sent and received are not time-sensitive.
The next issue manufacturing has faced in implementing Ethernet networks concerns how these networks suffer from latency issues %%MDASSML%% or what is referred to as jitter, in reference to the amount of time it takes data to be transmitted from one point to another. Due to network jitter, latency is not predictable; thus, data that might take 10 seconds to be received in one instance could take 30 seconds in another.
The issues with determinism, network latency and jitter are the largest concerns of applying Ethernet for the industrial control aspect of networking. This is because most control networks are designed to send small amounts of data through a network more frequently than Ethernet does. In addition, these control networks are often used for time-sensitive functions that require data to arrive in a controlled fashion.
Leveraging Ethernet benefits
Despite these issues, Ethernet offers several benefits when applied in a control environment. Despite the reduced frequency, Ethernet provides fast data transmission that proves advantageous in many manufacturing scenarios. As manufacturers require comprehensive, readily available information to make decisions quickly, the faster the necessary data can be passed to decision-makers, the quicker they can respond. Plus, Ethernet is inexpensive to apply, provides cost-saving advantages and offers lots of options for implementation in industrial plants.
One way that users can take advantage of Ethernet’s benefits is through distributed control, which takes network control and decentralizes it throughout the plant. With this arrangement, control is distributed among many controllers rather than one. In a traditional control network, a single master PLC is responsible for everything happening on the network.
The master PLC sends a command for some action to occur at another location. The receiver of this information obtains the request and performs the function that it has been instructed to do. In return, the receiver sends its response back to the master PLC indicating that the function has been completed. This standard implementation is exactly where the limitations of Ethernet come into play, as a time-sensitive request sent to a node on the network may be delayed due to network traffic: the more traffic (requests) on the network, the slower the node will respond.
Distributed control takes the responsibility of control from the main PLC to a decentralized location. In this way, each programmable gateway acts as its own controller, taking responsibility for the devices connected to it and passing along the data to a central PLC. The gateway reports to the master PLC, but the control is handled locally, which relieves the master of the full data load and reduces the amount of traffic on the network.
Another benefit users receive from this method is that the gateway will keep functioning even if the connection to the master is lost, so that the functions for which the gateway is responsible continue operating. When the connection to the PLC is restored, the data is automatically retrieved during the next scan of the system.
In addition to a distributed control network, another option for effectively applying Ethernet in manufacturing environments is to implement a subnetwork for control, with an Ethernet layer available for transferring the non-critical data. The programmable gateways used to facilitate this arrangement are designed to use a device level network below the Ethernet network and to manage local devices. At the same time, each gateway acts as the master on the subnetwork and a slave on the Ethernet network, making the data available to the Ethernet network.
These gateways allow users to bridge the gap between a control layer and a data layer. This is a viable option for users that have invested in a fieldbus infrastructure but would like to migrate to an Ethernet network. By using these gateways, the control layer remains on a fieldbus network, but the data are made available as if it is on an Ethernet platform. Consequently, users may still access information about manufacturing operations from the control network, enterprise intranet or anywhere the Internet is available %%MDASSML%% allowing users to take full advantage of the benefits Ethernet offers.
A distributed I/O system can be located near plant floor devices and used to monitor manufacturing systems, without requiring an additional enclosure.
Compact programmable gateways handle control locally. This gateway controls the temperature in a printing machine.