Industrial Ethernet signal clarity

Best practices and latest technologies help ensure the information received over industrial Ethernet networks represents the desired measurement or instructions. These includes proper grounding, cable routing, proper shielding, and cable length.

By Helge Hornis, PhD March 5, 2012

Ethernet protocols securely transmit data from source to sink. “Securely” in this context means that data sent by A and received by B is either:

  • Identified as uncorrupted
  • Clearly identified as corrupted such that repeat mechanisms take over, attempting to resend the data and get an uncorrupt copy to B.

It should be understood that Ethernet protocol designers tried to protect that data as much as possible. The Ethernet frame (level 2 of the OSI 7-layer model) uses a 32-bit CRC (cycle redundancy check) that is so hard to trick that the reception of unidentified but still corrupt data is so unlikely that it can be ignored. As a consequence, it is up to the installer and user to not undo these efforts through poor installation. In situations where plant noise can interfere with the data on the wire, several things can happen.

  • If the occurrence of the interference is infrequent, the data will eventually reach B undisturbed. Unfortunately, the requirement to repeat the transmission delays the arrival of data.
  • In another situation, A may never be able to successfully transmit the data to B before it gets a new update. When that happens, the data arriving at B is “jumping” more than it would under better circumstances, another possible consequence of delayed reception.
  • The third and the worst-case scenario is A simply cannot transmit any data to B. Still, users of well-designed communication systems can be quite confident that any data they are getting is good. In other words, the “network signal clarity” is a given.

Once one identifies bad installation practices as the main reason for data arriving at B late or not at all, it is helpful to know the elements of good installation practices. 

Proper grounding

Devices must be grounded properly. Many hardware manufacturers spend a considerable amount of time and effort to design grounding lugs and grounding terminals into their products. Use them. And don’t try to save a few pennies using a 24-gauge wire to ground the device. Since we are talking about networks, it is quite likely that the PLC is located relatively far from the data source. If so, it is a good idea to make sure that the entire machine is grounded to the same potential. When that is not the case, a shielded network cable may be selected to bring those machine sections to the same electrical potential, but this type of cable was not intended to do the job of a heavy solid or braided wire cable that was designed to provide a low-resistance path from a device to an earth ground.

Instead, disjointed machine sections must be connected by heavy-duty grounding straps and connected to a solid earth ground connection. When a shielded cable is brought into a metal junction box, immediately split the shield off and connect it to the box at the location where the cable enters. This method ensures that noise riding on the shield will not be brought near the equipment in the junction box but will be diverted to ground right away. 

Proper cable routing

A typical machine has many different electrical consumers, and some of those, for instance motors and drives, tend to generate higher levels of electromagnetic noise. One of the most common mistakes is to route the motor cables in close proximity to network cables. Don’t do it. The rule of thumb is to separate power and controls by 4 in. to 20 in., and this is one of those cases where more really is better.

Proper hardware selection

Step one of a successful network implementation is to select hardware that is well designed. This includes the network infrastructure as well as all other components. Getting back to the drives example, an installation that uses a drive with less electromagnetic (EMC) emissions will inherently be less problematic than one using a drive that significantly pollutes the spectrum.

Shielded cable? Don’t be misled

Many engineers want to go the extra mile and specify shielded cordsets. The only problem is that most shielded cordsets are useless! For a shielded cordset to work, the shield must somehow be connected to the machine ground. Unfortunately, in typical shielded cordsets, the shield simply extends along the length of the cable but does not connect to the connector or coupling nuts. The only advantage of this kind of cordset is that it is mechanically stronger. Any external noise “riding along the shield” will not be diverted to ground and away from the signal wires. Instead, the noise signal will have the opportunity to jump right on the signal leads at the connector.

Keep cables short

Any cable can act like an antenna. Keep it short. Also, do not help noise get onto your signals by looping long sections of unnecessary cable along the machine. This turns the cable into a transformer, making it easier for unwanted signals to couple into the line.

Free help

When the signal cable is placed inside of an open wire duct, push it into the corner. That way, both the bottom and side of the duct help shield the cable from external noise.


In situations where the noise comes in the form of high-energy pulses, ferrites are a good idea. For best performance, the network cable should be looped around the ferrite several times. Be sure to do this at each end of the cable also. Better yet, ferrites should be used on those cables that create the noise in the first place. Your PC monitor’s power cord has such a ferrite built in for a good reason.

When not to shield

While Ethernet cable should definitively be shielded and grounded properly, this is not necessarily the case for the AS-Interface network working in conjunction with Ethernet. This network is designed to be used with unshielded cable and, very importantly, neither of the two leads should be grounded. While shielding is possible and possibly useful in the worst electromagnetically polluted environments, using shielded cable will result in a 30% reduction of possible segment length.

See "Online extra" information below.

Hornis is manager, Intelligent Systems Group, Pepperl+Fuchs. Edited by Mark T. Hoske, content manager, Control Engineering, CFE Media, mhoske(at) 

ONLINE extra

Automation networks: Physical layer consolidation; Ethernet protocol discord

While the world of “automation networks” is getting simpler, this statement needs further explanation and amendment. It is true that Ethernet gets most of the attention and most professionals agree that it will be the technology of the future, but that does not mean we will have to worry about only one Ethernet-based networking technology.

As soon as Ethernet gained a level of acceptance, the battle shifted. Major PLC manufacturers each developed their own flavor of an “open” Ethernet protocol, effectively agreeing on a cable but not much more. Profinet does not talk to EtherNet/IP, and neither talks to EtherCAT. Unfortunately for the user, not all that much has been gained, but there has been some forward progre

  • Since PCs come with Ethernet ports, configuring PLCs that use these Ethernet-based networks has gotten easier. It is not necessary to have a dedicated hardware adapter converting RS232 to the network of choice.
  • Most Ethernet devices support some sort of setup and/or diagnostics via a web page. This is nice, as web browsers can be turned into visualization and diagnostics tools.
  • Chances are (but this is not guaranteed) that the physical network infrastructure (cable, switches, and connectors) can be used for any flavor of Ethernet solution. This is helpful for machine builders who find themselves developing for multiple PLCs on the market.
  • The installation rules for Ethernet-based solutions are independent of the protocol. This helps because good installation practices for Profinet will also be good installation practices for EtherNet/IP.
  • Ethernet protocols securely transmit data from source to sink.

This advice also applies to AS-Interface, the widely used low-level I/O network “under” Ethernet. More specifically, assuming a bit error rate of 10-4 (that is, on average 1 in 10,000 data bits is corrupted by an external influence), an AS-Interface system running 24 hours a day, 365 days a year will have one undetected error in roughly 2300 years, clearly a level that can be neglected.