Common Mistakes in PROFINET Network Design

Design mistakes that can cause intermittent PROFINET dropouts: office switches, rings without MRP, mixed traffic, faulty grounding and naming. The mistakes and how to fix them.
Technical summary
Technology
PROFINET RT/IRT, MRP, industrial switches
Topic
Industrial Ethernet design
Level
Intermediate
Prerequisites
  • Line topology and device list
In this article8 sections

The line stops for a few seconds twice a day. The PLC diagnostic buffer shows “IO device failed” followed by “IO device returned”; the operator restarts the line and the problem disappears until the next shift. These are the most time-consuming faults in PROFINET networks. Their source is often a decision made at the design stage.

PROFINET provides real-time communication over industrial Ethernet. Because it looks physically like Ethernet, people try to build it with office network logic; intermittent stoppages, unexplained communication errors and difficult fault finding often stem from this. The common mistakes are listed below.

1. Building the line with office switches

Unmanaged office switches do not prioritise PROFINET traffic, do not provide diagnostic information (port status, error counters) and are not resistant to industrial conditions (vibration, temperature, electromagnetic noise). Using PROFINET-compatible managed industrial switches is a basic requirement for both operational stability and fault finding.

PROFINET RT frames are marked as high priority using the priority field (PCP) in the VLAN tag. A switch that ignores this priority makes an RT frame wait behind a file transfer at a busy moment. If the delay exceeds the watchdog time, the controller considers the device disconnected. Another shortcoming of office switches is LLDP support: PROFINET uses LLDP for topology information and neighbour detection. Without this information, replacing a failed device without a programming tool (automatic naming based on topology) does not work.

2. Growing the topology without a plan

Networks where each machine added to the line is connected to “the nearest port” become impossible to trace over time. Very long line topologies can cause all devices behind a failed device to drop off. The topology should be planned on a drawing, and redundancy with a ring and MRP should be considered for critical lines.

In a line topology, the two-port switch inside each device acts as a bridge to the next device. When a device loses power or is removed for maintenance, all the stations behind it drop off. Each device added in series also increases latency a little; on long lines, the update time must be adjusted to this accumulation. The engineering software’s topology editor does this calculation when used.

A common mistake when building a ring is joining the two ends without configuring MRP. A ring without an MRP manager is, as far as Ethernet is concerned, a simple loop; a broadcast storm locks up the whole cell within seconds. In the ring, one device must be the MRP manager and the others must be in the client role.

After MRP has been configured, the recovery time must be compared with the devices’ watchdog time. MRP is defined in IEC 62439-2, and reconfiguring the ring after a break can take as long as the time defined by the profile used (at most 200 ms in the profile commonly used with PROFINET). If the watchdog time is shorter than this recovery time, IO devices will still drop off while the ring recovers. Redundancy has been set up, yet the outage still happens. If bumpless switchover is needed, options such as the watchdog setting or MRPD should be assessed according to the controller and device manufacturer’s documentation.

3. Mixing IT traffic into the PROFINET network

Carrying camera video, file transfers or office traffic over the same network can delay real-time traffic. PROFINET cells should be kept in their own segments; connections to other networks should be made through a controlled transition point (firewall or router).

IP cameras are particularly insidious. A camera stream using multicast is flooded to all ports on a switch without IGMP snooping configured, filling the limited network interfaces of PROFINET devices. If a camera really is needed in the cell, it should be kept in a separate VLAN, and multicast management should be configured explicitly on the switch. We covered how to draw the cell boundary in the article IT/OT network separation with the Purdue model.

Another common scenario is leaving the engineering computer permanently connected to the cell network. While this computer downloads updates, takes backups or scans network shares, it generates unwanted traffic in the PROFINET cell. If programming access is needed, a controlled service port or a defined flow through the cell firewall should be preferred.

4. Wrong cables and connectors

  • Using fixed installation cable instead of flexible (drag-chain) cable in moving applications
  • Connectors unsuitable for field conditions (IP rating, vibration)
  • Data cables running in the same duct as power cables without respecting separation distances
  • Incorrect shielding and grounding

These mistakes usually produce intermittent faults that are hard to reproduce; cable testing and visual inspection are the cheapest diagnostic methods.

Grounding the shield at both ends is the method recommended in PROFINET installation guidelines; however, for this to work properly, a proper equipotential bonding connection is needed between the two ends. Without it, equalising current flows through the shield and turns it into a source of noise. The real cause of dropouts that look “unexplained” on the network side is often an M12 coupling nut that has not been fully tightened, or a cable crushed at the bending point of a robot arm. Copper cable must also not exceed the 100-metre limit per segment.

5. Lack of device naming and documentation

PROFINET devices are identified to the controller by device name, not by IP address. Not giving a replaced device the same name, or using similar names, delays recommissioning after a fault. Device names, IP addresses, switch ports and topology should be kept in an up-to-date document.

Name assignment is done with the DCP protocol, and this can be triggered by anyone who can reach every device on the network. Assigning a name to the wrong device during maintenance leads to two devices carrying the same name and the line seeing one of them at random. The name of a drive replaced on the night shift should be entered into the managed documentation, not on a piece of paper. Keeping devices on the spare parts shelf at factory settings (unnamed) also prevents the same confusion.

6. Not using diagnostic data

Managed switches and PROFINET devices provide information about port errors, link drops and topology changes. Not monitoring this information means problems are only noticed when there is downtime. Industrial network monitoring makes it possible to see rising error counts before a dropout occurs.

For intermittent dropouts, the first place to look in the console is the CRC and alignment error counters of the relevant switch port. If the counter rises steadily, the problem is most likely physical: cable, connector, grounding. If the counter is clean and dropouts continue, look at traffic load, naming conflicts or the watchdog setting. Resetting the counters and reading them again one shift later shows which link is generating errors. For choosing the switch itself, see the industrial switch selection guide.

Checklist

  • Topology drawing and device list are up to date
  • PROFINET-compatible managed industrial switches are used
  • Redundancy (MRP) has been considered for critical lines
  • PROFINET cells are separated from other traffic
  • Cables, connectors and grounding suit field conditions
  • Port and device diagnostic information is monitored
  • MRP recovery time has been compared with watchdog times
  • Spare devices are kept unnamed and name changes are entered into the documentation

The primary source for installation guidelines and design directives is the PROFIBUS & PROFINET International (PI) website.

Frequently asked questions

Can an ordinary office switch be used in a PROFINET network?

It is not recommended except for short tests. Office switches are generally inadequate in terms of PROFINET prioritisation, diagnostic information and resistance to industrial environments. A PROFINET-compatible managed industrial switch should be preferred.

What is MRP?

Media Redundancy Protocol is a redundancy protocol that keeps traffic flowing in the other direction around a ring topology when a link breaks.

What causes intermittent communication dropouts?

Common causes are damaged cables and connectors, incorrect grounding, unforeseen heavy traffic on the network and incorrect device naming. Port error counters and PROFINET diagnostic information are the first places to look.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Industrial

PLC, PROFINET/PROFIBUS and SCADA communication, IT/OT network segmentation and industrial field connectivity. We install and monitor without stopping the production line.

Knowledge Center

Other articles