Problem in S7-300 AI module or in WinCC computer...

Guys,

I have really weird problem and hope some of you'll be able to help.
Basically, there is one S7-300 PLC (CPU 315-2DP). There are two IMs with couple
of Di and DO as well as AI modules. There are also two redundant PCs with
WinCC installed. One PC is Master, other is Slave. Both computer visualise
that signals becomes interrupted for 1 or 2 sec, and then it comes back to normal.
Time diagram of signal is shown on attached image. I thought that interrupts comes from
bad connections and I have tighten screws on both terminals and PLC's module connector.

Then I thought that maybe, there is some problem with pressure transmitter.
I opened SImatic Manager, found PIW variable and observed how it was changing.
There is classical Scale block that scales integer number to real engineering unit (bar).
I wasn't able to notice any interrupts (going vale ot zero) at all. Also, that measurement is
shown on WinCC graphics (I/O field), and it was present there without any interruptions. Refresh rate is 250 ms.
Interrupts can be seen only on trends.
Then I decided to disconnect transmitter and connect potentiometer (2 wire module configuration)
on the first terminal strip in the same cabinet in which LC is located.
Again I was not anle to see any changes in STEP7 (Online Monitoring), but again, there were random interrupts
on trends.
Then I opened trends on both PCs and noticed that both diagrams on trends containes interrupts but different
number and at different times, like they were different process variables.

I also wasn't able to notice any changes in mA (4-20) on my digital instrument connected in series with potentiometer.
I don't have any kind of oscilloscope to make further tests.
Is it possible that these interrupts are not real, but somehow false shown in trends, because they can only be spotted
on trends and nowhere else.
I mean, if interrupt is really present and last 1 sec, I think I'd be able to notice it
in Step 7 Online monitoring, wouldn't I?
Also if interrupts really exists, then they would look same on both trends, since data is from one PLC, right?

Has anybody had similar problems?

Thank you

http://www.plctalk.net/qanda/uploads/Signal.jpg

Pandiani said:

I mean, if interrupt is really present and last 1 sec, I think I'd be able to notice it in Step 7 Online monitoring, wouldn't I?

Also if interrupts really exists, then they would look same on both trends, since data is from one PLC, right?

Click to expand...

I'd say yes & yes.

Have you read this FAQ? Sporadic losses when archiving values

Hmm, that is interesting, I'll try to find out way to be sure this is the case.
I cannot exaplain that this phenomen is happening always on same variable. On trends, there are 4 or 5 process variables and problem exists only on one...
I'll try to find more tomorrow...