EtherNet/IP Performance Considerations

There are several areas to consider with regard to EtherNet/IP Performance:

General Recommendations

The following recommendations will improve the performance and determinism of the network in general.

Considerations for RMC EtherNet/IP Performance

The RMC controllers support the following EtherNet/IP I/O bandwidth:

Controller

Total Bandwidth (packets/second)

I/O Connections1

Minimum RPIs and Bandwidth

RMC75E

1250

1

2 ms (1000 pps)

2

3 ms (1000 pps)

3-4

3-4 ms (1000-1250 pps)

RMC150E

1250

1

2 ms (1000 pps)

2

3 ms (1000 pps)

3-4

4 ms (1000-1250 pps)

RMC200 CPU20L

6000

1-3

1 ms (2000-6000 pps)

4

4 at 2 ms (4000 pps), or

2 at 2 ms and 2 at 1 ms (6000 pps)

RMC200 CPU40

8000

1-4

1 ms (2000-8000 pps)

1 The RMC75E and RMC150E only support multiple I/O connections in multicast mode. The RMC200 supports multiple I/O connections in multicast or unicast mode. The first I/O connection always requires 2 packets per RPI. Subsequent multicast connections require 1 packet per RPI and subsequent unicast connection require 2 packets per RPI.

Considerations for Master EtherNet/IP Performance

The PLC EtherNet/IP I/O module must often support a large number of devices, including RMCs and other devices. Therefore, it is important to recognize the limits in bandwidth of your EtherNet/IP master controller, and to know how to determine the amount of bandwidth being utilized. This entire subject is covered in more detail in Rockwell Automation’s EtherNet/IP Performance application guide (Publication ENET-AP001D-EN-P), but will be addressed narrowly here for RMC connections.

Here are some published bandwidth limits for EtherNet/IP I/O controllers available as of this writing:

Manufacturer

Module

Total Bandwidth (packets/second)

Allen-Bradley

1756-ENET/B (obsolete)

900

Allen-Bradley

1756-ENBT

5000

Allen-Bradley

1756-EN2T,

1756-EN2TR,

1756-EN2F,

1756-EN2TXT,

1756-EN3TR,

10000

Allen-Bradley

1768-ENBT

5000

Allen-Bradley

1769-L23E

2000

Allen-Bradley

1769-L3xE

4000

Allen-Bradley

1788-ENBT

4000

Schneider Electric

Quantum 140 NOC 711 00

7500

Schneider Electric

Premium TSX ETC 100

7500

Omron

CS1W-EIP21

6000

Omron

CJ1W-EIP21

6000

Omron

CJ2H-CPUxx-EIP

6000

Omron

CJ2M-CPU3x

3000

Notice that it is recommended that not more than 90% of this bandwidth be used on I/O connections.

The bandwidth required for each RMC depends on the RPI and the number of connections. The first connection requires 2 / RPI packets per second (where RPI is in seconds), and each additional connection requires 1 / RPI additional packets per second. The following chart summarizes the requirement:

Number of Connections

Multicast Required for Bandwidth (packets/second)

Required Bandwidth for Unicast (RMC200 Only) (packets/second)

1

2 / RPI

2 / RPI

2

3 / RPI

4 / RPI

3

4 / RPI

6 / RPI

4

5 / RPI

8 / RPI

 

Therefore, to determine the total bandwidth required for any group of RMCs being serviced by a single master EtherNet/IP I/O controller, simply add up the bandwidth requirement of each RMC connection. Usually the RPI for all RMCs will be set the same, so the total utilization can be calculated as shown:

Total Required Bandwidth = ( # of RMC connections ) x ( 2 / RPI )

Example

For 12 RMCs with an RPI of 20 ms each and one connection each, the total required bandwidth is 12 x ( 2 / 0.020 ), which is 1200 packets/second.

 

Using this formula, we can also calculate the maximum number of RMCs that can be supported by your controlling EtherNet/IP module:

Maximum Supported RMCs = Total Bandwidth x 90% / ( 2 / RPI )

Example

How may RMC connections can the 1756-EN2T support at a 10 ms RPI? The Total Bandwidth for the 1756-EN2T is 10000 packets/second, so the maximum number of RMC connections is found by 10000 x 90% / ( 2 / 0.010 ), which is 45 RMCs.

 

If the total bandwidth required exceeds 90% of the master EtherNet/IP I/O controller's bandwidth, then you will have to consider one of the following options:

Example

Suppose one ControlLogix 1756-ENBT module will be controlling twenty-five (25) RMC connections. The intended RPI is 10.0 ms. Therefore the total bandwidth in packets/second is computed as follows:

Packets/Second

=

Number of RMC connections x (2 / RPI)

 

=

25 x (2 / 0.010s)

 

=

5000

 

The maximum allowable I/O load on the ENBT is 90% of 5000, which is 4500. Since 5000 is greater than 4500 packets/second, one of the following alternatives must be considered:

Considerations for Network Performance

The maximum amount of information that 100 Mbps Ethernet can carry in one direction is 100 million bits per second. If all packets on the network were EtherNet/IP I/O packets of the largest size supported by the RMC (125 four-byte REALs), then this would be a maximum of 21,600 packets/second (notice that packet overhead such as headers and inter-frame spacing are included in this estimate). The network utilization should be kept well below this maximum. As the utilization approaches the maximum, it becomes more and more likely that a packet will have to wait to get onto the wire, introducing small delays.

In addition to the limits of the wire itself, many devices have a maximum sustained rate that they can receive packets without falling behind in packet processing and eventually dropping packets. The RMC is one such device, since priority must be given to controlling motion above handling packets. No pre-determined maximum rate is currently available, but by reviewing the Ethernet Statistics in the devices, you can look to see if packets are being discarded.

Notice that EtherNet/IP multicast I/O connections can greatly impact the performance of an Ethernet network since—unlike unicast packets, which are sent directly to a single device—multicast packets are sent to all devices on the network by default. This increases the load on every device, including the switches, and each individual network segment.

If your network has a high utilization and/or network components are dropping packets due to high packet rates, you will want to look at ways of reducing the network utilization. There are three ways to reduce the network utilization:

The current way of doing this uses a protocol called IGMP (Internet Group Management Protocol). Using IGMP to direct packets in switches to only the interested ports requires that (1) the switch supports IGMP Snooping, and (2) at least one switch or router connected to the network supports IGMP Query. Notice that some switches are now available that provide both IGMP Snooping and IGMP Query in the same switch.

 

See Also

EtherNet/IP Overview


Send comments on this topic.

Copyright © 2025 Delta Computer Systems, Inc. dba Delta Motion