Hi...
we use one set of controllogix 5555 redundant system with controlnet connected to I/O chassis and ethernet 1756-ENBT card mount on remote chassis, and three computer (run xp sp1) using wonderware version 8.0 with 5000 tags connected via hub to contrologix. RSLinx 2.42 use to setup FastDDE (OPC) for wonderware. We have massage on rslinx opc event shows "controllogix got disconnected and re-initialize".
We set the RSLinx opc with polled message 500 ms and packet limit maximum 20000. We have problem with some data slowdown such as to open control valve, etc.
How to setup opc correctly on rslinx?
Thank you.
james

Are all 3 computers using their own RSLinx or you have just one RSLinx IO server?
You may be overloading. Some work around would be to reduce the number of IO servers and to use multiple topics. Use slower polling topics for stuff like setpoints, analogs, and data that are not critical. Use faster polling topics for the stuff you need to see change right away.

yes, its computer has its own rslinx professional version 2.42. We built only one topics for all I/O tags. Any idea?
Thank you

I would try disconnecting the hub and using just one computer, then see if you notice a better network performance. Sometimes Ethernet Hubs do not work well in this type of application. I would recommend using an Ethernet Switch instead...

JJ

If we run only one computer, network performance much better then two or three. Could the data traffic problem or incorrect poll rate setting / maximum packet limit setting?
james

If I had to guess...? all of your settings in RSLinx are probably fine, I think you are just expecting a little too much retrieving 5000 tags from the contrologix processsor on three seperate workstations.

I would recommend using one workstation as your primary OPC server and retrieve the data to the other workstations via Remote OPC from that worksation.

You may also want to try putting your E-Net card in the same rack as your Processor.....

JJ

I agree with TechJunki. You will get a lot better performance using 1 IO server vs. 3. Try the multiple topics idea and make sure Intouch is configure to poll only active items and not all items unless you need to update all 5000 tags at once for whatever reason.
Also you can increase the System Overhead Timeslice in the Contrologix. Be careful with this because it may improve your update time but at the cost of your program performance. If you have time critical logic, don't change this too high.

For our application, we want to keep all computers in same level.
If I want to add another topic to schedule the 5000 tags, say add 10 topics with each topic with diffrence settings (polled rate and maximum packets), separety the fast tag up date (pv - fast update and sp slow update), pressure, flow, any other critical items for fast update and slow update for non critical items.
Could this way will improve network performance? Any idea?
Thank you.
James

If you are determined to keep your curent network structure, reconfiguring multiple topics and updates will definetley be the first step to improving your problems...

You are polling a large amount of data, even with the separation of topics and slower update rates for less critical items, the performace may still not be up-to-par so to speak. Give it a try.... let us know how it turns out......


JJ