AuthorJames Powell

Frame Prioritization within Ethernet/IP and PROFINET?

Have you ever laid up at night wondering how Ethernet/IP prioritizes their messages or even if they ever did? Well maybe I was working too hard on writing my Ethernet/IP training course, or maybe I am a bit of a geek. Probably both.

At this time, I was reading a really good book on Ethernet/IP called ‘Ethernet/IP: The Everyman’s Guide to The Most Widely Used Manufacturing Protocol’ by John Rinaldi and Jamin Wendorf. There is not a lot of good information on this protocol, and this new book is certainly a welcomed addition to the literature. However, it did not say how frame prioritizing was done or even if it was.

How, or if, the prioritization was done has always bugged me as a communications engineer. I had read that Ethernet/IP could use VLAN but near as I could tell this was not widely used. So, how did they do it? I could not imagine that they had not incorporated a method for frame prioritization! If they did not have a method, then regular TCP traffic could delay IO type messages and who would design a network like that!

Being in the internet age, I decided to email John Rinaldi and ask him. His co-writer Jamin Wendorf had the answer; EtherNet/IP using Quality of Service (QoS) for Prioritization. The method of marking frames can either be VLAN tagging (802.1Q) or Differentiated Services (DiffServ – DSCP). The specification requires the use of DSCP, so that is what most vendors use. You can optionally support the VLAN tagging, but since most vendors don’t support it, it hasn’t gained traction in the field.

This answer lead me on an investigation that clarified the use of two terms QofS vs, CoS which I was confused with. QofS stands for Quality of Service and is an umbrella term for traffic prioritization. Class of Service (CoS) refers to a specific QofS as implemented in layer-2 Ethernet as defined in 802.1Q.

CoS is a 3-bit field called Priority Code Point (PCP) within an Ethernet frame when using VLAN Tagging as defined in 802.1Q. This all occurs at Layer 2 of the Ethernet frame.

This is how PROFINET uses Quality of Service in VLAN to prioritize messages.

Ethernet/IP also uses Quality of Service but not at layer 2; it is using Differentiated Services Code Point (DiffServ – DSCP) at layer three as part of the IPV4 header.

Assuming that your switch supports this QofS, then you are good for packet prioritization within Ethernet/IP.

Now, that this has been settled, I can sleep at night. As for which is a better way, I don’t see an advantage of one over the other as long as the switch supports the method you are using! The more I work with Ethernet, the more I realize that it always comes down to picking a good switch for your application!

If you want to know more about this topic, please consider taking one of JCOM Automation Inc. or Real Time Automation training courses.

 

What do extra terminations look like on PROFIBUS PA and DP and how can they occur?

What do extra terminations look like?

In both PROFIBUS DP and PA, having more than two terminations on a segment will lower the signal level. This lowers the overall robustness of the network and should be avoided.

In the case of PROFIBUS PA, the signal level is normally around 800 mV peak-to-peak.  Adding an additional termination will drop the signal level to around 600 mV, as shown below. This lower signal level can occur naturally if the network is very long. However, if the network is not very long and you see this, then you probably have an extra termination somewhere on the segment.

In PROFIBUS DP, the signal levels will vary from device to device and there can be multiple reasons for low signal levels; loose terminals, bad connectors or the device may just have low signal levels naturally. However, adding a termination will lower the signal levels for all devices on the segment. Below is an example where the regular signal level would be between +3 V and +4 V. Adding a third termination brings this down to close to +2V which is low.

 

How does this occur:

Anyone who has worked with PROFIBUS PA, knows that having more than two terminators is relatively common and certainly easy to do. Since many PA junction boxes have a termination switch or a place for a terminator, turning on more than the recommended two is easy.

PROFIBUS DP is a different story. When you turn on the terminator on the DB-9 connector, it will cut off the rest of the network. This makes it hard to have more than two terminations. However, just because something is hard has not stopped someone from doing it. The two ways that I have seen this done are:

  1.  A few devices have a termination switch on the PROFIBUS interface card. The customer may inadvertently turn on the termination switch on the card and on the DB-9 connector.
  2.  The DB-9 connectors are designed so that you extend the network by adding a wire to the ‘out’ of the connector. Some DB-9 connectors have a piggy pack connector that is supposed to be used only to connect diagnostic equipment. However, if you extend your network by plugging a second DB-9 connector onto the first one, then you have a setup where you can have double termination.

Conclusion:

Extra terminations lower the signal levels and this makes for a less robust system. Now, that you know how they can occur and what it looks like, you can avoid having this happen on your system.

What tools to use when troubleshooting Industrial Ethernet in general and PROFINET in particular?

There are many tools on the market for troubleshooting Ethernet and even ones made specifically for Industrial Ethernet. However, they vary greatly in functions and weather you use them on the network for a short time or if they are on the network all of the time.

For our Certified PROFINET Engineering course, I generated the following chart:

I listed all the typical issues that I would run into when commissioning a PROFINET network. The Permanent monitor that I was using was PROCENTEC’s Atlas and the Temporary monitor was PROCENTEC’s Netilities.

In truth, you can find all these errors with Wireshark. However, several of them, all of them that I have marked with a ‘No’, are very hard to find with Wireshark. The main problem being too much data and too hard to go through the data to find the issue. This is where tools like Atlas and Netilities shine. They give you more of a birds-eye view of what is going on.

One surprise I had was how good Atlas was at finding these issues. Also, how it is the only tool that I know of that will find cabling issues. Atlas does this by using some low-level Ethernet commands to ‘test out’ each segment. For industrial automation, this is very important since unlike the office environment, lost packets are a big deal.

There is a new hand held device on the market called Mercury which has the Atlas and Netilities software on it. You can also load Wireshark on it, so it can be your complete troubleshooting tool for Industrial Ethernet.

Office Cat-5e Cable in Industrial Ethernet Networks (Ethernet/IP, PROFINET, Modbus TCP)?

Industry, at least, has figured out that using office grade switches in industrial settings is a really bad idea. Those who decided to try office grade switches soon figure out it is not a great idea after they have replaced 2 or 3 in a year. Cabling is not that obvious. Anyone who has spent some time looking at Industrial Ethernet Networks has seen PLC’s wired with standard office grade cables into an industrial grade switch. They do work, or at least appear to work. Before I elaborate on that let’s look at what should be used.

Industrial Ethernet networks such as PROFINET recommend a shielded cable that is grounded at both ends. This grounding method is your best protection against EMC noise interference (see IEC 61000-5-2 for full details on why). Standard Office Cat-5e cable is un-shielded and therefore is more susceptible to EMC noise. Therefore, when you use standard office Cat-5e cable in an industrial setting you can expect to see at least some bit errors depending on the overall noise.

However, when talking to the people dealing with these networks and you ask them about bit errors, they say ‘We have not had a problem’. I can think of two reasons for this. First, Ethernet is so fast and the warning levels set high enough that you can have several problems but a valid message will still make it to the device in time. Second, these errors are still recorded but most systems are set up so that they are not visible. They are hidden away in the switches.

Therefore, to the casual observer, there are no problems. However, put a monitor like Atlas or Mercury on the network and you will see it.

The next questions are; Is this really a problem? Is it worth the added cost of using the recommended cabling?

I would say ‘yes’ and ‘yes’ because being a comms guy. I want rock solid communications. However, I am sure some of my customers will look at the added cost and say, no the added robustness is not worth it. At least they will think that right up until the network goes down.

If you would like help with your Ethernet Installation, please contact us.

Contact JCOM Today!

© 2019 JCOM Automation

Theme by Anders NorénUp ↑