Logix 5000 Trend Locks When Machine Fails

[kamenges]

I am working on a dual spindle winder controlled with a 1756-L71 PLC communicating to four Parker/SSD AC890 drives over Ethernet/IP. The drive speed generation and the tension loops are located in the plc. While running we get occurrences where we will see significant tension deviations, both high and low.

I set up a trend looking at spindle speed command and tension feedback on both rewind spindles. Here is the odd part. In every case that I have been trending these items, the trend had stopped updating at the same time that the tension issue occurs. Actually, the trend must stop updating slightly before the on-machine event since I never see the tension deviations in the trend. It isn't a long duration issue but at 1500 FPM it doesn't take long to push through enough material to cause some serious heartache.

It just seems too coincidental that EVERY time this happens the trend freezes. I have disconnected my laptop from the plc and the issue still occurs so I know it is not strictly related to the trend function or the laptop connection.

Has anyone ever seen anything like this before?

Keith

 

[T Gibbs]

possibly losing comms to the drives when it happens?

 

[kamenges]

In case there is any doubt in anyone's mind, I'm an idiot. I was causing my own problem.

I set up my own internal PLC data logger, pumping the items I was interested in into a set of parallel arrays. In addition to the items of interest I logged the routine call rate based on the difference in the processor current time value. I looked at the logged data after we had an event. Both of my rewind spindle speed command values took a 10% or so step down between two 10 msec scans and continued on as normal from there. When I looked at the routine call rate array at the same index as the speed step I saw that the time difference was about -60 seconds.

Here is where it gets fun. This system uses a ControlLogix PLC with a Factory Talk SE Local HMI. We have a recipe structure that is set up and held in the plc. One of the items in the recipe element is the date and time the recipe was last saved. This timestamp is based on the plc time. I wanted to keep the HMI and PLC time synchronizes so that the recipe save timestamp means something. There isn't an easy way (that I know of) to set a computers time to a plcs time. So I decided to go the other way. I read the HMI time from the plc and if they are different I use a SSV to set to plc time to the HMI PC time. We generate all the drive speed commands from the plc and send them over Ethernet/IP to the drives. We use a virtual motion axis to provide the base machine speed using a motion axis jog command. What do you think happens to active motion command when the plc time makes a 60 second step change? You guessed it; the instructions produce REALLY odd outputs.

A side affect is that the Logix trends appear to lock up. If I were writing the trend function I would use the plc time as my trigger and would set my trigger point as one sample interval into the future from the time at the sample I am actively taking. I suspect that is what Rockwell does. So when the plc time takes a jump backward it takes a much longer tome to reach the new trigger time. If I have been patient and waited long enough I suspect the trend would have started running again, at least for a little while.

So, there you have it. I'm an idiot.

Keith