Table of Contents
What is PPS?
In an office network, the key metric is Mbps (megabits per second). In an industrial network with 150 nodes, that is irrelevant. The metric that matters is PPS (Packets Per Second): how many packets the PLC's communications card can open, process, and close per second.
Every time a drive sends data to the PLC, it generates a packet. The card (such as the Omron CJ1W-EIP21) has a dedicated microprocessor. Every received packet is an interrupt for that micro. When packets arrive faster than the micro can handle them, the following chain reaction occurs:
- The card's buffer fills up.
- The card starts dropping packets.
- The PLC interprets the absence of packets from a drive as a lost connection.
- A network Timeout is triggered: emergency stop.
The real equation: how to calculate it
In EtherNet/IP, communication is cyclic and bidirectional. The drive sends data to the PLC and the PLC sends setpoints to the drive. Therefore, each connection consumes 2 packets per cycle.
The formula for the total packet load is:
PPStotal = ∑ ( 1000 / RPI × 2 )
Or simplified for a single node:
PPSnode = 2000 / RPI
| RPI (ms) | PPS per node | Notes |
|---|---|---|
| 10 ms | 200 PPS | Maximum load — only for critical sensors |
| 50 ms | 40 PPS | 80% less than at 10 ms |
| 100 ms | 20 PPS | 90% less than at 10 ms — standard for drives |
Key conclusion: dropping the RPI from 10 ms to 100 ms reduces card stress by 90%, with no real impact on the process for most devices.
Configuration in Omron CJ1W-EIP21
In the Omron ecosystem, management is done exclusively via the Network Configurator for EtherNet/IP. There are three critical parameters:
- RPI (Requested Packet Interval)
- Update frequency, configurable individually per connection (Target). It is the parameter that most influences total PPS.
- Connection Type
-
Multi-cast: the packet is replicated to all switch ports. Poison for large networks without managed switches.
Point-to-Point (Unicast): traffic goes directly to the destination. Mandatory in networks with more than ~20 nodes. - Load monitoring
- In the software go to View → Usage Chart to see real-time PPS consumption and compare it against the card's limit.
Official datasheet: CJ1W-EIP21 / EIP21S — Datasheet (Omron EU)
Hardware limits and the Domino Effect
Not all card versions are equal. Exceeding the limit causes total system collapse:
| Model / Version | PPS Limit | Overload symptom |
|---|---|---|
| CJ1W-EIP21 v2.1 or lower | 6,000 PPS | Random faults, card in "Error", drives stopped |
| CJ1W-EIP21 v3.0 or higher | 18,000 PPS | Greater stability, but the physical limit still exists |
The Domino Effect: when the PPS limit is exceeded, Jitter appears (variation in packet arrival time). If a packet should arrive every 50 ms but arrives at 200 ms because the card CPU was busy processing the other 149 nodes, the drive's watchdog interprets a cable cut and triggers an emergency stop due to a communications fault.
Golden rule for the field
Senior engineering is not about giving every node the fastest RPI; it is about giving each piece of data the speed it actually needs:
| Device | Recommended RPI | PPS per node |
|---|---|---|
| Safety sensor | 20 ms | 100 PPS |
| Motor / pump drive | 100 ms | 20 PPS |
| Flow meter / analog sensor | 500 ms | 4 PPS |
With 150 drives at 100 ms each, the total load would be 3,000 PPS — well below the EIP21 v2.1 limit (6,000 PPS) and far from the v3.0 limit (18,000 PPS). The network will run stably and drives will never interpret a communication loss.