Weird repeating Siemens issue

Lately one of our production lines has been randomly going down. It is all Siemens (S7-300), and about twelve years old, mostly original. The master CPU is a 315-2 DP, the communications processor is a 343-5, and the main HMI is a Simatic OP37/PRO. When it goes down, it gives the error message "No S7 communication peer found" on the HMI, and basically all the LEDs on the CPU and the CP are lit or flashing (I could specify which ones if anyone is interested).

I am new to Siemens, but by playing with the program (not connected to my running line) I have found a diagnostic buffer that I am hoping will lead me to an answer.

I have 2 questions..

1. Does it seem more likely that this is a communications problem, or is the communication error more likely a symptom of a hard fault on the PLC?

2. Anything in particular I should look for while diagnosing? Also, are there any special steps to changing out a module (say, a CPU) or do you just swap it?

Thanks everyone!

So, u have found diag buffer. What does it say?

Well, I haven't had a chance to check it while online with the machine.. it says it needs the PLC to be in stop, which is hard to do while it is running. Right now I am waiting for either a planned or unplanned shutdown.

jmc013 said:

Well, I haven't had a chance to check it while online with the machine.. it says it needs the PLC to be in stop, which is hard to do while it is running. Right now I am waiting for either a planned or unplanned shutdown.

Click to expand...

You're down the wrong path.
A dignostics buffer will not ask you to put the PLC in stop.
Maybe you're checking out "stacks" or something?
The buffer will display the current status with all the errors/entries but you have to be online.
If you highlight one of the entries and double click on it it will pop up a quick remedey/help menu on the fault.
Or is it display help I'm going by memory here.

So I can check the diagnostic buffer while the machine is running?
Follow up question: Does this seem like a reasonable way to determine the cause of this issue?

Yes (*2).

New problem: I can't plug in to the MPI port while running due to the fact that something is already plugged in there. There isn't another way to go online with the PLC is there? Also, do I have to plug in to the master CPU, or will any in the system work?

What is plugged in there? Dont it have piggyback connector? Is DP (the right side) port free?

jmc013 said:

New problem: I can't plug in to the MPI port while running due to the fact that something is already plugged in there. There isn't another way to go online with the PLC is there? Also, do I have to plug in to the master CPU, or will any in the system work?

Click to expand...

They sell profibus connectors with the build-in additional port in the back.
You can plug in directly to the back of the above connector without having to disconnect anything.
Do you have a profibus network?
Going on line with Profibus instead of MPI can be an option too.
Though, if you have a running system I would advice to be vary careful pluggin in your PC.
If you do go online with the profibus (not MPI) I believe plugging your laptop to any of the nodes will allow you to connect to the CPU.
A layout would be great, we are kind of assuming a lot as it is now.

Ok, I was able to get to the diagnostics buffer while the line was running and I saved the info out of the buffer for the master CPU and the CP. Just wondering if there is a way to save a screenshot of the profibus network? Also can I diagnose slave PLCs? I couldn't find the same information on those. I will post all I've collected after I figure out how to get this layout in a readable format.

You can diagnose the slaves by checking the hardware diagnostics of the master.
In STEP7 go to PLC ... hardware diagnostics (not the same as the diagnostics buffer).
This will give you a simple indication if the slaves are not available or have differences to the configuration. The error description will be a bit cryptic and sometimes may require that you think like a german to understand what they mean.

For a more in-depth analysis of profibus problems you need a dedicated profibus analyzer such as Profitrace. This is relatively expensive but worth it for pinpointing more tricky problems.

Here is what I have so far. The diagnostic buffer for the CPU (the dates and times are obviously wrong) shows the first error when the line went down this morning as "Cycle time exceeded," which I guess led to the following error.

The watchdog time is 150ms, which is what was exceeded, by the looks of it.

The buffer for the CP is not as helpful to me, as there are no times associated with the errors so I don't know when they happened. I am also trying to post the hardware configuration so you can see the jist of what I am looking at, but I have only figured out how to export it as a .cfg file, which I have no idea how to open. As soon as I can figure out how to get a screenshot I will happily post it.

JesperMP said:

You can diagnose the slaves by checking the hardware diagnostics of the master.
In STEP7 go to PLC ... hardware diagnostics (not the same as the diagnostics buffer).
This will give you a simple indication if the slaves are not available or have differences to the configuration. The error description will be a bit cryptic and sometimes may require that you think like a german to understand what they mean.

Click to expand...

This would only really work when the system is actually faulted though, right? I checked it just now but everything is running normally so there were no errors.

That is correct. The hardware diagnostics only shows entries when there are errors.

You may have to check the diagnostics buffer for older entries. The information in the diagnostics buffer may take some interpretation in order to figure out which DP node is causing the diagnostic entry.

JesperMP said:

You may have to check the diagnostics buffer for older entries. The information in the diagnostics buffer may take some interpretation in order to figure out which DP node is causing the diagnostic entry.

Click to expand...

Could you possibly look at the .txt files I posted? The interpretation is where I am getting hung up, I think. The first error is the "Cycle time exceeded." Any ideas on what could possibly cause this to happen?