Opinion from expert needed - Ethernet network - urgent!

Hi all!

I have to establish communication in a network with 9 nodes of S7-1200 and PC running WinCC Advanced.

Each node has S7-1200 with local KP400 HMI. This local data exchange is 70bytes or 40 tags for each node.

Each node has to communicate (have HMI connection) with WinCC exchanging 100 to 300 bytes (70 to 200 tags) each.

Between PLCs in nodes, i got also communication:
Node A gets 10bytes and puts 10bytes to node C.
Node B gets 10bytes and puts 10bytes to node C.
Node C gets 10bytes and puts 10bytes to node D.
Node D gets 10bytes and puts 10bytes to node E.
Node E is Profinet IO controler to a ProfinetIO devices node F and node G (35 bytes exchange each).
Node F gets 10bytes and puts 10bytes to node H.
Node H is Profinet IO controler to a ProfinetIO device node I (35 bytes exchange).

TIA portal V13 allows me to create such network and all connections.
Manuals says that for most distant node I (from WinCC) frame delay for 7 switches is total 0,35sec max.

But,

Is it possible to realise?
What about frame collison?
Would I have communication timeout for WinCC HMI connections, PUT/GET, loacl HMI connections, Profinet communication?

Thanks!

Anyone?

I implemented a similar scenario with 8 S7-1200 PLCs and 3 Siemens large HMIs. Its a simpler version of your architecture, I think. No WinCC though. Worked great in the end (no collisions or timeouts) but was difficult and time consuming to implement, test and debug.

So I believe your scenario is possible using just PUT/GETs. It will require a lot of thought and testing. I don't think you'll have a problem with collisions or timeouts when you are done. However, the limited communications resources of the S7-1200 and KP400 will be your most likely problem: you need to make sure each HMI and PLC does not exceed its maximum number of connections or you will get strange, intermittent behavior. The TIA portal does not tell you when you've exceeded the maximum connection resources so it requires very careful design.

When using the S7-1200 PUT/GETs, you'll want to consider chaining them to run serially on each PLC to minimize simultaneous connections. If you run them serially (rather than in parallel) by waiting for the prior PUT/GET to finish before starting the next one, the S7-1200 can talk to many other S7-1200's.

You'll also want to ensure your PUT/GET is designed to ensure it does not hang up if a connection dies. If you do all your PUT/GETs in series, then any failure of a PLC will cause a chain of timeouts/hangs in the rest of the series PUT/GETS. But if you do them in parallel, you run out of connection resources. Tricky stuff.

Maybe you could you come up with a 'simpler' architecture where one PLC is the 'master' that GET/PUTs everything between PLCs?

I also found it was simpler to understand by changing my architecture so everything was done with PUTs. No GETs. So any PLC that had a change would PUT to one master PLC using a distinct communication DB for that exchange. That master PLC would then PUT each communication DB to each appropriate slave PLCs using the exact same DB. By separating the comm exchanges into distinct DBs, you had a chance to follow it all in the code.

I'm sorry my answer is probably too vague to be of direct use to you (no example code). Though possible, you'll find this fairly time consuming to implement, test and debug IMHO. We went to the trouble and time because we used this system over and over at multiple sites.

Siemens has an example here.
Still studying it to see if it will help with my application.

Thanks.

I must NOt have a master (head) PLC because its a water supply system witth 4 groups inside.
Pumps Node A and B pump to reservoir node C.
Reservoir Node C supply reservoir node D.
Pumps in node E supply water to reservoirs node F and node G.
Resrovir node F supply pumps in node G, and
pumps from node G supply reservoir H.

So, if PLC node A fails (no mains, error, broken etc)... reservoirs and pumps C to H are still full and operational so those subsystems can and MUST work.

TIA portal does calculate communication resources (see attached snapshot) once put/get command is properly configured, but i decided to use profinet because Profinet does not enter communication resources budget. Profinet uses less CPU comm resources, but uses more ethernet/network resources; that's reason I needed help and opinion

I can use PUT/(GET) in series, because PLC will be syhncronised via SNTP and i can program PUT(A to C) on 0, 3, 6, 9sec; PUT(B to C) on 1, 4, 7, 10sec...

With all stations selected, and compiled, info window shows NO ERRORS NO WARNINGS.

Another option may be to use I-device for your S7-1200 stations

steffen_sohl,

I already have 3 of PLCs in role of IO-device here.

HOMER_BL,

Thanks for the further explanation. You have an interesting and fun application there.

Here is the quote from the General Specifications and Features for the S7-1200 that bit me in my application (really, seriously buried in the S7-1200 documentation):


That '3 connections for server' and '3 connections for HMI' was not shown anywhere on the screens: only the 8 client. When I exceeded the 'server' connections, I got no compile errors and the system would work for a few hours, then communications would croak. So I re-architected to limit the server and HMI connections and the communications became fully reliable.

I may be dating myself here: this was with V3 hardware and V12 TIA Portal. Maybe this has changed and no longer a potential problem. I don't even know if that spec I quoted above is still true - but it sure bit me in the past.

nwboson,

I am not worried abput S7-1200 specification, I am worried about ethernet network. Then "longest" connection passes seven switches - sholud WinCC be able to exchange data in 1 sec interval or will I get "####" in data fileds on PC-screen? For Scallance, store-forward is 50ms (including optic to electric transition) for 256bytes frame - which gives delay of 0,35sec in one way, and 0,7 sec for two-way (command confirmation). And, switch in node A deals with all WinCC traffic - data for all nodes passes this switch.