Dispelling myths about industrial Ethernet, part 2

There are still a lot myths about industrial Ethernet floating around out there.

There are still a lot myths about industrial Ethernet floating around out there.

In the first part of our series on the most common myths about industrial Ethernet, we looked at a few widespread misconceptions about Ethernet solutions in the factory, including the notion that popular protocols such as PROFINET, EtherNet/IP et al are not really deterministic. Much of the confusion about Ethernet's place in automation and control systems is rooted in Ethernet's original status – which lives on its office-grade flavors – as a probabilistic network that is increasingly prone to collision as utilization rises. Plus, unlike the serial fieldbuses and point-to-point 420 mA connections that once dominated industrial automation, Ethernet-based networks carry a diverse range of traffic, not just time sensitive data intended for real-time applications.

Industrial Ethernet protocols, network management techniques like segmentation (via managed switches) and supplementary technologies such as PriorityChannel, with its combination of a custom CPU and advanced filtering, have more than addressed these inherent drawbacks in Ethernet [1]. Still, several myths live on what Ethernet can and cannot do in industrial settings. Let's dive into a few more of them and get down to the truth of the matter. Having covered three big myths in part one – pertaining to non-determinism, wireless solutions and the speed-throughput relationship – we will start with number four and go from there

Myth #4: An industrial Ethernet protocol monopolizes access to the associated Ethernet port
Like many other myths in this space, this one is the result of applying the logic of fieldbus to Ethernet. In the heyday of serial communications, it was true that running something like Modbus would prevent other protocols from accessing the same port [2].

The same doesn't hold for Ethernet, which as we noted above is built to accommodate many applications at once. I/O from PROFINET, EtherNet/IP, Modbus TCP and other standards may be run alongside applications carrying HMI data destined for a printer, or HTTP and FTP traffic for Web sessions.

Moreover, this versatility within Ethernet makes multiprotocol support the rule rather than the exception. The barrier to adding another application is just not high to bypass the benefit of having multiple solutions at your disposal.

Hence the importance of innovations such as the Real-time Ethernet Multiprotocol switch at the heart of the RapID platform [3]. It takes care of real-time protocols including SERCOS III, Ethernet Powerlink, etc., in addition to handling topology management via beacon-based protocols as well as the upcoming specifications from the IEEE Ethernet groups.

Myth #5: Industrial-grade cabling and other equipment are nice to have, but not really necessary
It should be clear enough that industrial Ethernet diverges significantly from its office counterpart when it comes to performance and reliability. Industrial-grade protocols are also very different animals in terms of what types of equipment they need in order to function optimally.

Essentially, commercial cabling won't always cut it. The common environments for industrial Ethernet include factories, power plants and refineries in which there are high levels of noise (both electrical and electromechanical), vibration and shock, on top of additional possible interference from corrosive gases and extreme temperatures.

Control cabinets may not always be an option for shielding Ethernet infrastructure from harm. Industrial-grade cabling and hardened switches are essential to success under these challenging conditions.

Myth #6: PROFINET relies on non-standard Ethernet and also does not use the TCP/IP stack
Of all the major industrial protocols, PROFINET is almost certainly the subject of the most myths. This is likely due to its popularity and the way that PROFINET IRT devices use switches for high levels of performance.

In reality, PROFINET utilizes standard Ethernet, which can be discovered by looking at a sample frame using a sniffer [4]. The PROFINET data is there, contained in an IEEE 802.3-compliant frame. It does not require a proprietary chip.

"A standard Ethernet controller chip can be used for PROFINET, even for real-time data," explained Carl Henning of PI International in one of the organization's official blog posts [5]. "Just add a PROFINET stack (the software that knows how to direct the PROFINET message as well as TCP/IP traffic). In order to achieve motion control, PROFINET does require that the stack be implemented in silicon; that is, in a chip. Otherwise the required speed cannot be achieved. However, the chip is not proprietary."

PROFINET is also an open technology, with numerous stacks, modules and chips available from companies that have bought rights to it from IEC. Compared to the competition, it is relatively easy to implement and offers a wide range of compatible solutions.

Confusion also surrounds PROFINET's use of TCP/IP. It actually does put TCP/IP to work, but only in specific contexts, namely communications about diagnostics and configuration. It does not depend on TCP/IP for real-time messaging, however, since that particular software stacks adds uncertainty and delay to delivery.

Instead, PROFINET skips TCP/IP and taps into EtherType 8892 for direct relay of the message to the PROFINET application. The ability of standard Ethernet to distinguish between different traffic types has made it especially useful for these types of use cases, in which real-time data must be distinguished from the rest of the noise across the network. PROFINET itself and accompanying solutions such as PriorityChannel capitalize on this feature to ensure ideal performance of real-time applications.

Myth #7: The higher specified cable frequency is always better
Ethernet provides much more bandwidth to utilize than any fieldbus, and as such, it can support a much larger number of network devices. However, the extra bandwidth is not always needed and may end up costing plant operators a premium in unnecessary high-frequency cabling.

The low baud rate of serial connections is sufficient for some less demanding applications [6]. In other cases, Cat5e 2-pair twisted cabling capable of 100 Mbps data transmission rates is ideal – Cat6 (for Gigabit Ethernet) may be overkill in terms of the applicable speeds and the cost of purchase and implementation [7]. Note that mixing Cat5e, Cat6 and other cable types and connectors on the same network is usually not recommended since it can cause performance degradation [8]. 

[1] www.acromag.com/sites/default/files/White_Paper_What_Is_PriorityChannel.pdf

[2] www.rtaautomation.com/newsletter/Newsletter4.html#3

[3] www.innovasic.com/technology/rapid-platform-faq

[4] us.profinet.com/4-profinet-myths-much-better-realities/

[5] Ibid.

[6] See Note 2

[7] www.cablewholesale.com/support/technical_articles/megahertz_madness.php

[8] web.arrownac.com/sites/default/files/pdfs/Selecting_Ethernet_Cabling_and_Connectors.pdf

Comments are closed.