I went to a customer site a few days ago to help solve a problem involving PROFINET and failsafe. In the scene accidentally found a PROFINET network interference phenomenon, here to share with you.
At that time, when checking the switch status information, found that there are switch statistics show that one of the ports will have an error message, such as the following figure of port 2, there will be error packet statistics, and the number will slowly increase. But the site connected to port 2 did not drop off, nor did it have any other problems, so the customer did not pay attention to the site. Therefore, at the beginning to the scene to see the phenomenon and did not have in-depth inspection, and so on to deal with customer concerns, before turning back to continue to deal with.
Ethernet port statistics data show that there is an error message even if it is a CRC error. Generally, the data is disturbed in the transmission process, and a test error occurs. Although it does not drop the station, it is risky to the network. Upon inspection, the site connected by port 2 and the switch are in the same electronic control cabinet, the upper and lower two rows, the length of the network wire is not more than two meters, also using Siemens industrial Ethernet wire and metal connectors.
It is normal to use a special tool to check the wire, and no obvious errors have been found in the wire and connector. For further investigation, a number of substitution experiments were carried out.
Change the port on the switch, the fault is still there.
Replace the cable, the fault is still.
Replace the distributed IO site interface module, failure as before.
Change the firmware of the switch, the fault is still the same.
Later, with the help of multimeter detection, it was found that the peripheral metal part of the RJ45 interface of the distributed IO module is not connected with the installation backplane when not plugged into the network cable, and only when the Internet wire is plugged in, can the metal part be turned on.
This indicates that the equipotential connection between the switch and the IO site is realized through the shield layer of the network wire. This will cause when the two ends of the line potential is not equal, there will be a current through the shield, and then cause data line interference, data error. But why the module RJ45 interface metal periphery punches the part and the backplane does not open, originally is installs the guideway to cause. A metal shrapnel is connected to the installation guide rail on the back of the interface template, and the metal shrapnel is conductive to the peripheral metal part of the RJ45 interface. But there is a coating on the installation guide, which is insulated, which is very difficult to see from the naked eye, thinking that it is a common aluminum rail. There does seem to be a scepticism about exclusion, even where it is taken for granted.
In this case, although there will be an error message, it will not cause the station to drop. This is because the PNIO site has a watchdog time set for off-site decisions, and the default is to allow three lost PNIO cycles to be accepted, meaning that the watchdog time is three times the refresh time. This way, as long as it is not a continuous error packet, it will not cause the station to drop.