Calculating Ethernet I/P I/O RPI in Controllogix

[Chris Elston 78]

Last week, I swear I saw a PDF in the KB on how to calculate the RPI setting based on the number of I/O cards you have in your remote Ethernet I/P racks or something of that nature. This week, I can't seemed to find it. I looked in: EtherNet/IP Modules in Logix5000 Control Systems: ENET-UM001I-EN-P EtherNet/IP Performance: ENET-AP001D-EN-P There are bandwidth calculations BASED on RPI... but not the RPI setting. I thought there was a formula I saw. Or heck if anyone has a Excel cheat-sheet, that would be great. Where did I see this at? What manual? Or did I dream it up? Using a 1756-ENBT with a few of the 1734-AENT. All digital I/O sitting in the point I/O blocks, so it will be rack optimized to consolidate the connection.

[Ken Roach 194]

Was it specific to the 1734 POINT adapter ? Since the POINT adapter scans the POINTBus backplane like it was a little DeviceNet network, the number of I/O modules in the Rack Optimized connection will determine how fast you can get an update for all of them; any faster and you're oversampling. There's a nice Excel-based calculator for root bridge capacity; you tell it how many devices on how many adapters and it generates a bandwidth and CPU utilization estimate for the "scanner" root bridge module. It's not what you are asking about but it's a handy thing to have. http://www.rockwellautomation.com/solutions/integratedarchitecture/resources3.html#enetpredict

[paulengr 44]

Normally the RPI is your choice and the result (in packets per second) is calculated based on the RPI's. Set your RPI's based on three considerations: 1. The update rate of the I/O card. Setting RPI's any faster has no value since the input won't change any faster. 2. Set it to the lowest update rate demanded by the process. You may set it slower based on process requirements. For instance it's pretty much a waste of bandwidth to sample termperature inputs at a rate any faster than about once a second for almost all processes. And in some cases, you can use "COS" (change of state) mode on inputs that change very rarely but require a quick response such as operator push button panels. In this case the RPI is just there as a "heartbeat". 3. Pay attention to the network capacities (packets/second). For instance ENBT's, most AENT's, and such, have a capacity between 2500 and 5000 packets per second, while xx-COMM-E interface cards on Powerflex drives only have a capacity of 600 packets per second. Most Ethernet switches have a capacity of around 20,000 packets per second or more, so they can easily flood the individual IO cards. So keeping track of your total packet throughput is an important overall design parameter when setting your RPI's. The only calculation then is simply taking your RPI's and the number of connections and calculating the resulting overall packet rates. Since an RPI is in milliseconds, 1000/RPI = packets/second per connection.