Random Plant stoppage, I question DeviceNet network.

The last two days our automation has just stopped randomly at different spots. Everything that has stopped, is thru DeviceNet.We have about 5 faults on our HMI ( Panelview Plus )at the same time. Digging deeper into the faults, I took notice that they are faults dealing with a air cylinder ,reed switch, or motor not reaching a prox , etc. that has to reach that in a certain amount of time , or it will fault out.

I am thinking that the program is still running , but the DeviceNet Flex I/O is losing comunication for a split second.

I am not getting any DeviceNet faults. Do you think I am on the right track ? Have any of you had a DeviceNet network that has somewhat brokendown ,after being in service for a few years ?

Thanks in advance , have a great evening.

How are you detecting and trapping DeviceNet errors/faults ? What type of DeviceNet scanner are you using ?

DeviceNet uses a standard "4x RPI" timeout value. If you're using Polled I/O, and there are four Polls by the Master that are not responded to, the Master declares a timeout.

Similarly, if you are using Cyclic I/O connections, the Master declares a timeout after more than 4x the RPI has gone by without a reply from the slave.

After a Timeout has occurred, it takes between 1 and 4 seconds for a DeviceNet scanner and a FLEX adapter to re-start the connection. This is plenty of time for ladder logic to trap a fault bit, but too fast for most folks to check the scrolling display on the scanner.

It's certainly possible for a DeviceNet network to degrade over the course of a few years, just like any electronic system with field terminated wires. Corrosion, vibration, or increased noise on the system could start causing CAN network errors, and those could start causing I/O connection timeouts. Such problems absolutely will show up using a Woodhead NetMeter.

What's very unlikely is for the system to be able to have enough packet errors to miss I/O signals without also generating I/O Connection faults that can be trapped and diagnosed.

Is there anything else the I/O points have in common; a common DC supply or DC Common wire that might have loosened or failed ?

Be thankful the operators aren't describing this as a "PanelView fault" because the error messages are displayed on the PanelView.

I plan on purchasing a Woodhead NetMeter tomorrow. I read the manual a little before leaving the plant. I didn't see a discription of how to actually hook it up to the network. Is there any better discription on how to hook it up?

It seems like a difficult problem to solve ,being so random.

Thanks for your help,

The NetMeter comes with some tap cables that are meant for plugging into Mini and Micro quick-disconnects.

I frequently installed short drop cables for analyzer access, and made up my own cable with an open-style Combicon female socket. This would allow me to plug in either the NetMeter or my 1784-PCD for protocol analysis.

The NetMeter frequently does its job by showing no errors ! If I'm investigating a DeviceNet and it goes through a device failure but there are no CAN errors on the bus, I know that the device itself should be investigated, rather than the network.