Modbus TCP · EtherNet/IP · PROFINET · Switch hops + WAN round-trip latency
When to use: Use this to estimate the round-trip communication latency between a controller and a field device across an industrial Ethernet network. Total latency is built from protocol stack processing, per-switch store-and-forward delay, frame serialization, cable propagation, and any WAN link. This determines the fastest practical poll/update rate and whether the link is fast enough for closed-loop control versus supervisory monitoring.
OT (operational technology) network latency determines how quickly a SCADA master or PLC can exchange data with field devices, and it directly governs whether a communication link can support closed-loop control, supervisory monitoring, or only periodic data logging. This simulator decomposes round-trip latency into its components — protocol stack processing, switch store-and-forward delay, frame serialization, cable propagation, and WAN links — so engineers can evaluate protocol choices and network topologies before deployment.
Total round-trip latency (RTT) is the sum of one-way path latency multiplied by two, plus device processing time. One-way path latency = protocol stack processing time + (number of switches × store-and-forward delay) + frame serialization delay + cable propagation delay + WAN delay. Protocol stack overhead is lowest for PROFINET RT (approximately 0.25 ms, prioritized Layer 2) and highest for Modbus TCP (approximately 1.2 ms, TCP/IP overhead). Each managed industrial switch adds approximately 10 µs store-and-forward latency for small frames at GbE speeds.
Frame serialization delay = (frame size in bits) / (link speed in bps). A 64-byte frame at 100 Mbps takes (64×8) / 100×10⁶ = 5.12 µs. Cable propagation delay ≈ 5 ns/m in copper (signal velocity approximately 2×10⁸ m/s), so a 1000 m cable adds 5 µs. WAN links (cellular, satellite) add the largest latency — cellular 30–80 ms, geostationary satellite 500–600 ms one-way.
IEC 61784 defines fieldbus profiles for industrial Ethernet, including PROFINET, EtherNet/IP (CIP), and Modbus TCP. PROFINET RT (Real Time) is defined in IEC 61158-5-10 and achieves cycle times of 1–4 ms for standard I/O. PROFINET IRT (Isochronous Real Time) achieves cycle times as low as 31.25 µs for motion control, requiring special managed switches with hardware-based time synchronization (IEEE 802.1AS). IEC 62443 imposes cybersecurity requirements including monitoring for latency anomalies that may indicate traffic injection or DoS attacks on OT networks.
For closed-loop control, RTT must be well below the process time constant. A flow control loop with a 100 ms time constant can tolerate approximately 10–20 ms RTT; a temperature control loop with a 60-second time constant can tolerate 1–5 second RTT. PROFINET RT or EtherNet/IP Class 1 cyclic communication is appropriate for process I/O with cycle times of 1–10 ms. Modbus TCP is appropriate for supervisory polling of setpoints and diagnostics where 50–500 ms latency is acceptable.
Network segmentation via managed switches with VLAN tagging and QoS (IEEE 802.1p) should prioritize OT traffic over IT traffic on converged networks. Minimize the number of switch hops between PLCs and I/O: each hop adds 10–100 µs depending on switch technology. For wireless links (industrial Wi-Fi, cellular), add additional latency margin of 10–50 ms and design for occasional packet loss.
Select the industrial protocol (Modbus TCP, EtherNet/IP, or PROFINET RT) which sets the protocol stack baseline latency. Enter the number of switch hops between the master and the field device, the link speed (100 Mbps typical, 1 Gbps for backbone), total cable length in meters, and any WAN or router delay if the remote device is on a different network segment. The calculator outputs per-component latency, total one-way latency, round-trip time, and the maximum practical update rate (with a 1.5× margin for polling overhead).
At 100 Mbps with 3 switch hops and 300 m cable, Modbus TCP round-trip time is approximately 3–5 ms, giving a theoretical maximum of 200–333 requests/second. In practice, Modbus is polled at 1–20 Hz for SCADA monitoring because the bottleneck is the device response time (1–50 ms for most PLCs and RTUs), not the network. For fast I/O, use PROFINET or EtherNet/IP with deterministic cycle times.
PROFINET RT uses prioritized Layer-2 Ethernet frames with VLAN tag priority 6 (IEEE 802.1Q) to bypass the normal TCP/IP stack. The cycle is synchronized across all devices by a PROFINET controller, and each device transmits its I/O data in a fixed time slot within the cycle period. Standard RT achieves 1–4 ms cycles; IRT (Isochronous RT) adds hardware timestamping and IEEE 802.1AS clock synchronization to achieve 31.25 µs–4 ms cycles with jitter below 1 µs.
A managed industrial switch adds approximately 5–15 µs cut-through latency or 50–150 µs store-and-forward latency for small frames (64 bytes) at 100 Mbps. Cut-through switches begin forwarding as soon as the destination address is read; store-and-forward switches receive the entire frame before forwarding to check for errors. For low-latency applications, use cut-through managed switches (e.g., Cisco IE, Hirschmann, Moxa). At GbE, both modes converge toward 5–10 µs.
Cellular LTE links are commonly used for remote SCADA monitoring and supervisory control of slow processes (water/wastewater, pipelines, remote substations), but they are not suitable for closed-loop control due to 30–100 ms RTT and variable latency. Typical application: 5-second poll rate for process values, 1-minute setpoint updates. For critical control functions, a hardwired or licensed radio backup link should be maintained. Cellular also requires security hardening per IEC 62443 due to public network exposure.
RTT (round-trip time) is the time from when a Modbus request is sent to when the response is received — it is a network/device parameter. Poll cycle time is the interval at which the SCADA master sends a new request to the same device — it is a configuration parameter. Poll cycle time must be greater than RTT plus any device processing time. Setting poll cycle time equal to RTT leaves no margin and will cause missed responses when latency varies. A practical guideline is poll cycle time ≥ 1.5 × RTT.
Try our SCADA Studio
More calculators, simulators, and guides for this discipline.