Dueling Views of PROFINET’s Place in System Architecture

Posted by & filed under Uncategorized.

I was amused by these dueling viewpoints in recent Control Engineering articles:

  1. Ethernet for sensor networks? Why it makes sense today
  2. Consider a sensor network to ease connections

The first one argues for network-connected, machine-mounted IO that can be close to the sensors and actuators.  That’s a good argument and I fully agree with it.  There are many machine-mountable IO blocks with PROFINET (and PROFIBUS) connectivity.  It makes sense to group sensors and actuators inputs and outputs close to where they are, allowing a single cable run back to the controller.  And that cable run can make stops at other IO blocks near other clusters of sensors and actuators.  We actually demonstrate this in our PROFINET one-day training classes; look at the device circled in red (on one of our training racks):

Class Rack

The second article basically says Industrial Ethernet is bad – AS-interface is good.  This is just wrong – each has its place.  There are projects where clustering the IO as the first article advocates makes sense; there are other projects where the IO is spread out and AS-i makes sense.  I see AS-i as an occasional complement to PROFINET; it is definitely not a replacement for PROFINET.  In fact, AS-i is so much a complement to PROFINET that PROFIsafe can integrate AS-i safety.

There is another mistake in this second article which must be corrected.  In fact there are two errors in the section “Wasted resources.”  It says that PROFINET requires special hardware.  WRONG!  PROFINET can use standard Ethernet controllers with a PROFINET stack.  PROFINET can also utilize silicon to meet regular real-time IO needs (which is what AS-i does… and PROFIBUS, and DeviceNet, and FF, and most other fieldbuses.)  PROFINET can also utilize silicon to achieve greater levels of speed and determinism.

The second error in the article says that coexistence of Industrial Ethernets will no longer be possible.  Also wrong!  One of the great things about Ethernet is its ability to support multiple applications at the same time. Your email arrives even while you are surfing the web.  Modbus TCP, EtherNet/IP, and PROFINET can coexist on your infrastructure (cables and switches) with each other and with web servers in Ethernet switches and with other standard Ethernet traffic.

Now back from error-correction mode: When it comes to resolving these dueling viewpoints, my advice as always is: “Do the engineering.”  Evaluate the alternatives.  Remember that cost is a legitimate design parameter.  Study, if needed.  Then choose the architecture that is right for your project.  I’m convinced you will choose a PROFINET backbone complemented if needed by such other buses as the requirements may demand.  I even wrote an article about that for Control Engineering and Plant Engineering: Fieldbus in the field.

–Carl Henning

One Response to “Dueling Views of PROFINET’s Place in System Architecture”

  1. Brad Hart


    I have just been reading up on IPv6. Siemens had a white paper describing the differences between IPv4 and IPv6. The number of addresses for IPv6 is supposed to allow Ethernet to exist as a true one-to-one connection and eliminate NAT and PAT. With such an enormous range of addresses possible – 340,282,366,920,938,463,463,374,607,431,768,211,456 – I would believe it to be possible to have Ethernet sensors, utilizing PoE. For sensors/actuators that require more current, provide a secondary power source. Is this practical? Maybe. Utilizing the same principal of GSD/EDS files for I/O stations and smart devices would be applied to smart Ethernet sensors and actuators.

    There would be innovation again in Proximity and Photo Eye Sensors with sensor companies developing better diagnostics (taken from IO Link). An industrial protocol would have to take the lead on this and I believe PROFINET is in a position with the Device Name and Topology technologies to easily make this happen. It would be a much bigger challenge for EtherNet/IP and ModbusTCP to get there. Siemens has already made the innovation with IO Link in there contactors/motor starters for diagnostics, and this could be applied to IPv6.

    In place of Topology, you could easily do a browse for Industrial Ethernet devices, and those devices would appear in a list. A graphical/text indication would determine if that device has been spoken for in another system. Then, any unspoken Ethernet device could be linked/synced to the controller with its GSD file uploaded from the sensor (no more having to hunt down this file). Then, it is an easy association of symbol/tag name(s) to bit/byte address(es) of the Ethernet device. Once the Ethernet devices have been spoken for/owned with the PROFINET device name, then press a button to learn the Topology. Therefore, if a device is replaced, even if the device (i.e. prox) is a different brand, because of the standard interfacing of a prox, the new replacement device would take on the PROFINET devicen name and start working.

    Even Push Buttons and pilot lights could become smart Ethernet devices. The days of the I/O cards would be numbered.

    Just thinking!

    Brad Hart
    C&E Sales

« Back to PROFIblog