Hi all. TIA V16, CPU 1516-3 PN/DP, 14 ET200SP stations and two G120 on one PROFINET line. Thermoforming line we shipped last year, fine until about three weeks ago.
Diagnostics buffer is full of pairs like this, a different device each time:
"Distributed I/O: station failure. IO device: st-07"
"Distributed I/O: station return. IO device: st-07"
Sometimes st-03, sometimes the second G120, and the gap between failure and return is a second or two. I swapped the IM on st-07 with st-03 and the fault didn't follow it, and the update time is 2 ms everywhere so I tried 8 ms on the worst ones, no change. Every port link light is steady.
Is this a bad IM that only shows under load? Any ideas before I start swapping 14 of them.
Draw the line for me. Which port of which device goes where, and what does the second cable run connect to at each end? Segment lengths too.
Switch model, if any. And are the devices daisy chained through their own ports, or is there a switch in the middle?
If you mirror one port and see the same frame arrive twice in Wireshark, it's a loop, not a bad IM.
It's a plain line. CPU port 1 to st-01 port 1, st-01 port 2 to st-02 port 1 and so on down to st-14, the drives are in the middle. Longest hop about 40 m. Only switch is an unmanaged one on the CPU's second interface for the HMI.
The second run: st-14 port 2 back to CPU port 2. The electrician said the drawings called it redundancy. I didn't put that on the drawings, I left the topology at line and never looked again.
Tried 8 ms on all of them last night, still faulting.
Sorry if this is basic, what's MRP? Same thing as a ring?
MRP is the ring protocol. Your line is now a physical ring with no ring protocol, so it is just a loop. Frames go round, devices choke, one drops, comes back, the next one drops. That is your pairs.
Two ways out. Pull the st-14 to CPU cable and you are back on a line tonight. Or keep it and configure MRP: CPU interface, Advanced options, Media redundancy, role Manager, ring ports 1 and 2. Every device in the ring, same dialog, role Client, its two ports as ring ports. All of them, one missing and it is a loop again. Compile, download the lot.
Someone did a decent article on the ring versus line part: https://plctr.com/advanced-plc-networking-and-communication-protocols/
Pulled the second cable first thing this morning and the buffer's been quiet all shift. So it was the ring with no MRP, my topology said line and reality said ring.
Then did MRP properly over lunch: CPU as manager, ports 1 and 2 as ring ports, all 14 stations and both drives as clients. Reconnected the cable, downloaded, still quiet. Pulled a cable in the middle to test and nothing dropped, which is what the electrician was after all along.
Still odd: st-11 took about 20 seconds to come back after the download and the rest were instant. Didn't chase it, not sure it matters. Thanks dieter_k, and petek for putting the loop in my head before I went looking at the cabling.