ControlLogix - ProSoft Modbus TCP/IP - RPI Setting

[Eradicated]

I am not very experienced in Rockwell PLCs and have run into a small problem, so I could do with a little bit of help.

We have a redundant ControlLogix system (two PLC racks and one I/O rack), as per the screenshot below:

As you can see, we have two ProSoft Modbus modules (MVI56-MNET), which are used for supervisory-to-PLC communications.

The problem is that sometimes the Modbus messages sent from the supervisory are being rejected by the ProSoft card (i.e. we are receiving Modbus exception code 06 - Slave Device Busy).

According to some information found on the Internet and ProSoft website, this could be addressed by reducing the RPI for the module (currently set to 250ms for both Modbus modules in the I/O rack).
However, when one of our field engineers tried reducing it to ~50ms, seems like this started generating too much traffic and the Modules would eventually stop responding and fail (needed a restart).

The question is: is there any way to calculate minimum acceptable RPI for those modules? The ProSoft manual states that: "fifty milliseconds is usually a good starting value" for a module in a remote rack.

I would rather avoid trial-and-error approach if I can, but I cannot find any guidelines on calculating it?

 

[AlfredoQuintero]

I would advice you calculate the slowest RPI which your process can tolerate and use this number for yor setup. Why update at a faster rate than needed if the faster rate affects this one component of this network ?

 

[danw]

My Modbus experience is that determining the highest polling rate is an exercise in empirical experimentation. It is dependent on the components/devices in the system and network conditions (like EMI noise levels from external sources).


The only way one knows is to keep increasing the polling rate (lowering the interval time) until it causes problems.



I have read hundreds of OEM Modbus manuals and I have NEVER seen a calculation for maximum throughput, although I have read numerous recommendations on MINIMUM polling rates, especially for slave/servers.

 

[ASF]

According to some information found on the Internet and ProSoft website, this could be addressed by reducing the RPI for the module (currently set to 250ms for both Modbus modules in the I/O rack).
However, when one of our field engineers tried reducing it to ~50ms, seems like this started generating too much traffic and the Modules would eventually stop responding and fail (needed a restart).

I think you've misunderstood the advice (and it appears to have been poorly worded, leading to your confusion).

If the issue is that the module is too busy to process any more requests, you need to reduce the load on it. If it can perform one task per ms, and you give it 250 tasks to do, the absolute fastest it can do those tasks is 250ms. If you then tell it to do that same number of tasks in 50ms, it's going to have no chance - the problem will get much worse, as you discovered.

When they said "reduce the RPI", I expect what they meant was "reduce the polling frequency", which is the inverse of the RPI. You should increase your RPI value, which will decrease your polling frequency, and give the module a little bit more breathing space.

As others have suggested, work out how slow you can *actually* tolerate data update, and start there. If it's borderline at 250ms, I'd try 500ms - but you'd likely get away with 300-400ms if you needed it.