As many machine builders, OEMs, individual plants, and large corporations decide to move from the “bus” to the “net” (Profibus or DeviceNet to Profinet or EtherNet/IP) they have a chance to look at all the new architectures available and decide on which is the best for them. Here are the first two topics to take into consideration:
Topologies: With Ethernet based networking there are many to choose from: Star topology – point to point cordsets run from switch to switch or switch to device. Any cordset failure between the switch and device will not affect any other portion of the network. This topology is the most expensive, but provides ease of troubleshooting. Linear topology – seen quite often with Profibus and DeviceNet, allows the ability to daisy chain network products, which has the lowest cost, but is the most difficult to troubleshoot. A single cordset failure can take multiple Ethernet products offline. Each device in a linear topology must support an embedded switch, so care must be taken to ensure that all types of network traffic can pass through, especially safety protocols. Ring topology – just as it states, the network devices are daisy chained together with the last run of cable returning to the original starting point, the communication module, thus completing a ring. Most rings are set-up as a one fault tolerant system. One cordset break will not shut down your network. Allen Bradley’s DLR ring, boasts the ability to determine the two devices the cordset break is between. To receive this level of functionality the device must support the ring architecture, just because a device has an embedded switch does not guarantee support on a ring topology. On the downside, this functionality usually comes at a price, so you will pay extra for a one fault tolerant network.
Global Standard: Most flavors of Ethernet based networks boast a global standard, but the question you need to ask is how widespread is the standard accepted. Be sure to check all the global testing locations where devices are evaluated, usually if they are located in only one country, it may not be a good sign. The next thing I would check are the availability and variety of devices on the market, the trend setting networks usually have many devices from multiple vendors. In a nutshell, make sure you can buy similar devices from multiple vendors, so your network can survive cost increases by having other options.
Looking at Topologies and Global Standards is just the start, I will follow this post with other topics, including: Cost Effectiveness, Diagnostics, Scalability & Flexibility, and Trends in the market.
To learn more about industrial networks visit www.balluff.us
John, as you point out: it’s good to be cautious in applying linear topologies. Device failure or loss of power part way down the linear connection will cause the rest of the devices in that line to lose connectivity. If losing one device takes the production line down anyway, then no harm. If it does not take the rest of the production line down, then a star might be the right choice… or a star with some linear spurs. There’s never a substitute for just doing the engineering. You have probably seen the tips on topology we offer in our PROFINET one-day training classes: http://us.profibus.com/training.aspx?pagetype=oneday I’ll make it a point to blog about them at http://www.PROFIblog.com to make that information more widely known. I previously introduced the topic as “PROFINET is not a Star”. http://www.us.profibus.com/wordpress/?p=981
PROFINET also offers ring topologies. Two popular options are MRP (Media Redundancy Protocol) and our bumpless ring.