Hi all. Packaging line, 1756-L72S, two 1791ES-IB16 safety input blocks, safety RPI 10 ms on both. Ran three years without a safety connection drop.
IT swapped the unmanaged switch in the line cabinet for a managed one two weeks ago, and since then one of the two blocks drops its safety connection three or four times a day. Connection timed out, 16#0203 I think, inputs go to safe state and the line stops. The other 1791ES on the same switch has never dropped.
I tried the RPI at 5 ms instead of 10, thinking faster would help, and it got worse, so it's back to 10. Reseated the module, new patch lead, and IT says the switch shows no errors.
Is the switch faulty, or is it something on the safety connection? IT say the switch shows nothing, so I'm not sure what else to be looking at on the safety side.
Switch model first. And what did IT enable on it, IGMP snooping, storm control, anything QoS? Same VLAN as before or did they move ports? A managed switch left on defaults is not the same as an unmanaged one.
Stratix 5700, replaced a plain 8 port unmanaged. IT ran the express setup and left the defaults, whatever those are. Same VLAN, same ports as the old one by the labels.
The RPI change tells you a lot. The connection reaction time limit on a safety connection is derived from the RPI times a multiplier. Halve the RPI, halve the limit, more drops. So your limit was already sitting just above the minimum, and the new switch adds a little latency the old one did not. Measure before replacing anything, the switch is probably fine.
Module Properties on that 1791ES, Safety tab, Advanced. Connection Reaction Time Limit, and under it the Timeout Multiplier and the Network Delay Multiplier. The limit is built from the RPI and those two. Default timeout multiplier is 2 if I remember right, sitting there since commissioning with no margin.
RPI back to 10 ms, then raise the timeout multiplier a notch, 3 is where I'd start, and look at the network delay multiplier the same way. The CRTL goes up and the switch latency fits inside it.
Not on your own though. That number is part of the safety response time, so the stop distance gets redone, the safety task re-signed, and your safety engineer signs the change. What firmware is the L72S on?
RPI back to 10 ms, timeout multiplier 2 to 3, downloaded with the safety engineer watching. CRTL field went up, I have the before and after in the validation sheet. Six hours, no drop. Signature not re-recorded yet, he wants the response time sheet first.
Get IT to mirror that port for a day. Nothing dropped means latency, not loss.
Two days, no drops on either module. Safety engineer redid the stop time with the new CRTL, still inside the curtain distance, safety task re-signed, new signature in the file. IT mirrored the port for a day, no drops on the capture, so latency it is.
So the reaction time limit had no margin over the RPI minimum and the managed switch added enough latency to trip it. Timeout multiplier raised in Module Properties, Safety tab, Advanced, response time recalculated and signed.
The other 1791ES never dropped on the same switch and I don't know why. Firmware on the L72S is v31 for whoever asked. Thanks rsdave, boris_h and petek.
Faster RPI to fix a timeout. Everyone tries it once.