Despite the proven fact that industrial Ethernet networks are more capable and just as reliable as the fieldbus networks they are increasingly replacing, there are many people in industry who remain skeptical of the technology. Their skepticism is largely based on: factors that no longer impact the technology, how the technology was once applied without the latest network advancements, and plain old misinformation.
To clear up these misunderstandings, following is the truth behind the 10 most persistent Ethernet myths.
Industrial Ethernet is not deterministic
Technically, this is true … but it leads to an incorrect assumption. When switches replaced hubs, collisions were eliminated because the switch helped avoid potential collisions. Unlike hubs, switches detect potential collisions and delay one of the two packets by a fraction of a second so small (called “jitter”) that it has no effect on real-time factory communications. Theoretically, if you had enough traffic through enough switches on a network, and one device needed to communicate through all of those switches to control a device at the other end of the network in order to do precision motion control, you would have a problem. But today’s precision motion control networks avoid this in a variety of ways, with one of the most popular methods being IEEE-1588 Precision Time Protocol. This capability is embedded in motion controllers, as well as some industrial switches and servo drives to provide exceptional results.
Industrial Ethernet can’t do the toughest real-time control
When the world was full of hubs, 10 Mbit/s Ethernet and half-duplex communications, this myth was true. Today’s switches, however, with 100 Mbit/s and faster networks, and full-duplex communications being the norm, industrial Ethernet is successfully being used by millions of automation devices and in the toughest applications.
Increasing the Industrial Ethernet speed doesn’t do much to increase throughput
This was very true of older fieldbus networks, but not Ethernet. For example, increasing DeviceNet’s baud rate from 125k to 500k could slow down the network’s throughput. Or bumping an Allen-Bradley Remote I/O network from 57.6k to 230k would only improve scan times from 6ms to 9ms per drop. In both cases, factors other than the speed of the network were restricting communications. When you move from 100 Mbit/s to 1 Gb on Industrial Ethernet, however, you get a 10x improvement in performance. With today’s industrial Ethernet technology, network performance is often negligible in determining overall system throughput, with automation devices’ firmware and field devices’ processing time being the major contributors to speed.
Designing a network is hard/Designing a network is easy
The truth is actually somewhere in-between. The network design rules required for industrial Ethernet are easier to apply than most of the technology you use on a regular basis. But two pitfalls exist that make industrial Ethernet seem more complex than it is and which can limit the success of your design. First, industrial Ethernet inherited many of its features and options from commercial Ethernet. As a result, a lot of training courses and seminars present various theories and capabilities of Ethernet compiled over the past 30 years, even though many of these capabilities are no longer used. Second, designers often take shortcuts to reduce cost while, in exchange, they end up complicating what could be a very simple and easy-to-manage design.
If I keep my factory network separate from my enterprise network, my factory is secure
Many call this the “air gap”. There are plenty of reasons this is false: Google the terms “Stuxnet”, “flame”, “duku” and “zero-day vulnerability” just to read about a few. Then think about how well you can control USB jump drives from entering your space. The same goes for DVDs and CDs. These are all holes in your system that can be exploited. Any network in use today, no matter how it is architected, requires advanced security to effectively protect against breaches.
Wireless won’t work for real-time control
Standard wireless technology, specifically IEEE802.11n, has evolved so significantly from its predecessors — 802.11a, b, and g — that it now works well for the vast majority of real-time industrial applications. Most notably, 802.11n added: speed to reach 100 Mbit/s wire speeds, retries to emulate wired industrial network behavior, and the ability to make signal reflections add together to increase distance and coverage while older technologies produced dead-spots. While some features such as roaming, signal strength status checks, and certain security routines may interrupt communications, many can be adjusted or disabled to accommodate demanding performance.
Commercial-grade Ethernet cable will work just as well as industrial grade
While the control cabinet may protect commercial grade switches, cables and connectors from chemicals and abrasion, you should consider the full spectrum of industrial concerns, including:
- Electromechanical noise from all of the other devices in the cabinet
- Electrical line noise hitting the commercial grade power supply
- Shock and vibration affecting the power and network connectors, as well as the electronics
- Industrial approvals and certifications for your industry
- Extreme temperatures unless your cabinet is air conditioned
- Corrosive gases in the air
- Mean time between failure ratings of the device
Fiber is too hard to terminate
While adding fiber connectors once required very careful work and precision to get it right, today’s field terminations are very easy.
Always use shielded copper/Always use unshielded copper
There are three main rules to guide your consideration of shielded or unshielded copper, and their application revolves around the existence of electromechanical noise near your Ethernet cabling.
Rule 1: Always use bonded pair industrial Ethernet cable. Bonded pair significantly helps reduce noise, especially after installation when bending and pulling from installation causes gaps between the twists in non-bonded pair cables. These gaps allow for noise and other problems that reduce performance.
Rule 2: If you’re near very high power, high electrical noise devices always use shielded bonded-pair. If not, an unshielded bonded pair is fine.
Rule 3 – Always specify your cables and connectors. Otherwise your installer is likely to use commercial grade … see Rule 1.
Never mix video, voice and real-time control signals on the same network
For a very long time, managed switches have provided a feature to prioritize traffic through the switch called Quality of Service (QoS), which allows different forms of traffic to coexist and perform very well. While you might choose to keep these kinds of traffic separate, it isn’t usually necessary.