Hey all.
Filling line here. ControlLogix 1756-L75, one 1756-EN2T in slot 1, Studio 5000 v33. Ever since the historian went in last month the FactoryTalk View SE clients show stale tags and the operators get a comms error on the fill station display.
The EN2T webpage, Diagnostic Overview, has the CPU sat at about 100 percent and connections current 118. I think the limit is 128. Talking to that module there's four SE clients, the SE server, the historian, a Kepware box IT use, and three other controllers doing MSG reads into this one.
I/O never drops. I rebuilt the RSLinx Enterprise shortcut and rebooted the EN2T on Sunday and it was back at 100 percent inside an hour. Is the EN2T failing, or is 118 just too many? And if 118 is too many, what comes off first, the Kepware box or the MSG reads from the other three controllers?
What's the historian polling at, and how many tags did it add? Same question for the Kepware box, poll rate and tag count. Before we blame the module I want to know what the new load actually is. And those three MSGs from the other controllers, are they cached?
The historian's 1,400 tags at 1 s. Kepware is 300 tags at 500ms, and IT say it's "just monitoring". No idea what cached means on a MSG, I do HMI. Each SE client sits on about 900 tags on the busiest display, scan rate left at the default 1 s.
Cache Connections is a checkbox on the MSG instruction, Communication tab. Leave it unticked and every MSG opens a connection, does its read, closes it, every single trigger. Three controllers doing that at some rate is a steady churn of opens and closes, and the EN2T's CPU pays for all of it. Tick it on all three and the connection stays open, one each. That's the cheap part. The bigger part is that every class 3 client in the plant is sat on the one module that's also carrying the I removed link If I remember right the EN2T is good for around 128 TCP connections, and the CPU is what runs out first.
Cached the three MSGs, the other lines' programmers did it over lunch. Connections went from 118 down to 96. CPU's still 95 to 100. The SE clients still throw the error bar, a bit less often. Historian and Kepware are both still on there, so I guess that's most of it.
Deadband on the historian. And 1 s on 1,400 tags isn't monitoring, it's trending.
Deadband helps the historian, but the fix is a second module. Stick a 1756-EN2T, or an EN2TR, in a spare slot with its own address, move the SE server shortcut, the historian and the Kepware box over to it, and leave slot 1 for the I/O and the produced tags. Module properties, Port Configuration, give it the next free address, then in FactoryTalk Linx edit the shortcut path to the new module. The controller doesn't care which module a client comes in through, so there's no code to change. Slot 1 stays clean whatever the office does. Somebody did a decent article on the connection side: https://plctr.com/utilizing-ethernet-ip-in-plc-communication/
Second EN2T went in Saturday, SE server, historian and Kepware all moved onto it. Slot 1 is at 22 percent CPU and 31 connections. Clients have been clean for three shifts.
Everything HMI and SCADA was piling through the one I/O module. Cached MSGs helped, the second module fixed it.
60 percent on the new one still feels high to me, so deadband next. Thanks rsdave, scott_m.
60 percent on a module nobody's I/O depends on is fine. Deadband anyway.