What is Cycle Time?
Cycle time is the time an industrial controller, such as a PLC, needs to complete one full program cycle: reading inputs, executing the program and driving outputs. It determines how quickly a control system can respond to a change in the process and how predictably it does so. In real-time systems, a short and stable cycle time is the foundation of deterministic behaviour. In manufacturing, the term has a second meaning: the time a machine or line needs to produce one product, a key figure in OEE.
🧠 How does a single PLC cycle work?
A classic PLC runs a continuously repeating scan cycle. One cycle consists of four steps:
- Read inputs - the CPU copies the state of all digital and analogue inputs into a memory area, the process image of the inputs. For the rest of the cycle the program works with that snapshot, even if a sensor changes in the meantime.
- Execute the program - the logic is processed from top to bottom, written in IEC 61131-3 languages such as ladder diagram or structured text.
- Write outputs - the results are transferred from the process image of the outputs to the physical outputs or over the network to O.
- System tasks - communication with HMIs and SCADA, diagnostics, self-tests and any online changes.
The cycle time is the sum of these steps. Because the amount of code executed can vary from cycle to cycle (jumps, loops, conditional blocks), the cycle time of a free-running PLC is rarely completely constant. Engineering software therefore always shows a minimum, current and maximum cycle time.
🔄 What is the difference between a free cycle and a fixed cycle time?
The IEC 61131-3 standard organises program execution through tasks. A task can run periodically (with an interval) or start on an event, and it has a priority. Each vendor implements this in its own way:
| Execution model | How it works | Examples | Suitable for |
|---|---|---|---|
| Free-running cycle | A new cycle starts as soon as the previous one ends; cycle time varies with program load | OB1 on Siemens S7, continuous task on Rockwell Logix | General machine and line logic |
| Periodic task | Starts at a fixed interval, regardless of how quickly the previous run finished | Cyclic interrupt OBs (OB30 to OB38), periodic task, TwinCAT and CODESYS tasks | PID control loops, sampling, counters |
| Event-driven | Starts on a hardware signal or a software event | Hardware interrupt OB, event task | Fast response to one specific signal |
| Bus-synchronous | Task runs in step with the fieldbus cycle | Isochronous mode with ProfiNET IRT, EtherCAT distributed clocks | motion control |
A fixed cycle time has one major advantage: time-dependent calculations, such as an integrating controller or a speed measurement, get a constant time interval. In a free-running cycle the program has to measure the elapsed time itself.
🔧 How does a controller monitor its cycle time?
Every industrial controller has a watchdog: a monitor that intervenes when a cycle takes too long, for example because of an endless loop. What happens next depends on the platform:
- Siemens S7-1500 - in TIA Portal you set a maximum cycle time, 150 ms by default. When it is exceeded the CPU calls the time error OB80. If OB80 is not loaded, or if the maximum cycle time is exceeded a second time within the same cycle, the CPU goes to STOP. An optional minimum cycle time stretches short cycles so that the cycle becomes more even.
- Rockwell Automation Logix - every task has its own watchdog, 500 ms by default and adjustable from 1 ms to 2,000,000 ms. A time-out raises a major fault (type 6, code 1). It applies to the whole task, so a program fault routine does not catch it; only the controller fault handler can deal with it, otherwise the controller faults.
- TwinCAT and CODESYS - every task has its own interval and watchdog; the runtime counts how often a task overran its interval. For a soft PLC on an industrial PC, that counter is the main indicator that the operating system or other software is disturbing the real-time task.
A watchdog is a safety net, not a design limit. If you leave the maximum cycle time far above the normal value, you will notice late that the control system is getting slower.
🏭 Which cycle times are typical?
The right cycle time depends on the speed of the physical process. Typical ranges:
| Application | Typical cycle time | Explanation |
|---|---|---|
| Motion Control and servos | 0.1–4 ms | EtherCAT targets ≤ 100 µs; PROFINET IRT goes down to 31.25 µs |
| Fast machine logic (packaging, sorting) | 1–10 ms | Often a separate fast task next to the main cycle |
| Standard PLC control | 5–50 ms | Free-running cycle of an average machine or line |
| Process control in a DCS | 100 ms–1 s | Temperature, level and pressure change relatively slowly |
| Safety control (Safety PLC, SIS) | 10–100 ms | Driven by the required response time, not by the SIL level |
| SCADA polling and visualisation | 250 ms to several seconds | No real-time control; supervisory oversight |
Network cycles count too. PROFINET RT typically runs at cycles of 1 to 10 ms; IRT, using fragmentation and time synchronisation via PTCP, a protocol based on PTP (IEEE 1588), goes down to 31.25 µs. EtherCAT achieves synchronisation well within 1 µs using distributed clocks.
📏 How do you determine the right cycle time for your installation?
A practical approach in five steps:
- Identify the fastest event the control system must detect: a cam on a conveyor, a pulse from an encoder, a pressure surge.
- Calculate with the worst-case response time. If an input changes just after the PLC has read its inputs, the program only sees that change in the next cycle. The maximum response time is therefore roughly twice the cycle time, plus the filter time of the input card, the output delay and any network cycles.
- Translate it to the process. A product on a belt moving at 2 m/s travels, in the worst case, 2 × 20 ms × 2 m/s = 8 cm with a cycle time of 20 ms before the output switches. If the tolerance is 1 cm, you need a faster task or a hardware interrupt.
- Tune control loops. A rule of thumb is that a control loop should sample at least ten times faster than the time constant of the process. A temperature loop with a time constant of one minute can run at 1 s without problems; a pressure control on a pump cannot.
- Check safety functions. Under IEC 61511 and IEC 61508, a safety function must act within the process safety time: the time between the onset of a hazardous condition and the moment the accident becomes unavoidable. A widely used design rule is that the total response time of sensor, logic and actuator is at most half of that.
Then measure the actual maximum cycle time under full load, with all communication active, and keep a margin of 30 to 50 per cent for future extensions.
⚠️ What causes an unstable or excessive cycle time?
- Program load - loops over large arrays, string handling and compute-intensive functions in every cycle.
- Communication - many simultaneous HMI, SCADA or OPC UA connections consume CPU time in the system phase.
- Network problems - network congestion and high latency delay I/O updates and cause time-outs.
- Operating system - on a soft PLC without an RTOS or real-time extension, updates, virus scanners or virtualisation can interrupt the task.
- Online changes and diagnostics - downloading during operation and extensive tracing temporarily lengthen the cycle.
The visible consequence is jitter: variation in the cycle time. It leads to irregular control actions, missed pulses and poor synchronisation between axes. In the worst case the watchdog intervenes and the machine stops.
✅ How do you optimise cycle time?
| Measure | Explanation |
|---|---|
| Split the program across tasks | Fast logic in a short periodic task, slow logic (reporting, calculations) in a slower one |
| Do not run everything every cycle | Calculate slow values once per second, use edges and events |
| Separate communication | Route heavy connections through a communication processor or edge gateway |
| Use a real-time platform | An RTOS or real-time kernel gives a soft PLC stable timing |
| Monitor the cycle time | Log minimum, average and maximum values; trend them in the historian |
| Choose suitable hardware | A faster CPU or a separate motion controller offloads the main controller |
🏷️ What does cycle time mean in production and OEE?
Outside control engineering, cycle time means the time between the completion of two consecutive products on a machine or line. In OEE, the ideal cycle time determines the performance factor: (ideal cycle time × total count) ÷ run time. Three related terms are often confused:
| Term | Definition | Example |
|---|---|---|
| Cycle time (production) | Actual time per product on a machine or station | 42 s per housing |
| Takt time | Available production time divided by customer demand | 27,000 s ÷ 600 units = 45 s |
| Lead time | Total time from raw material to finished product, including waiting | 3 days |
If the production cycle time is longer than the takt time, the line cannot meet demand. PLC cycle time and production cycle time are related: a machine with a 10 ms control cycle can never switch more precisely than that interval, which limits the achievable product rate.
❓ Frequently asked questions
What is a normal cycle time for a PLC?
A normal cycle time for a PLC in typical machine control is between 5 and 50 milliseconds. Motion control requires cycles of 1 millisecond or less, while process control loops often run at 100 milliseconds to 1 second. The right value depends on the speed of the process.
What is the difference between cycle time and scan time?
Cycle time and scan time are used as synonyms in practice for the duration of one PLC cycle. Some vendors use scan time specifically for program execution and cycle time for the whole cycle including I/O and communication. Check your platform’s documentation to see which definition it uses.
What happens if the cycle time becomes too long?
If the cycle time of a PLC exceeds the configured maximum, the watchdog intervenes. Depending on the platform, the CPU calls a fault routine or goes to STOP, which puts the outputs into a safe state. Even before that limit, a long cycle time already causes slow responses and missed signals.
How do I calculate the response time of a PLC?
The maximum response time of a PLC is roughly twice the cycle time, plus the input filter time, the output delay and any network cycles. Two cycles are needed because a signal that changes just after the inputs are read is only processed in the next cycle. For critical functions, a faster task or an interrupt is the solution.
Does the SIL level affect the cycle time?
The SIL level does not directly determine the cycle time of a safety PLC. What matters is the process safety time: the total response time of sensor, logic and actuator must stay well within it. The SIL level mainly sets requirements for reliability, diagnostics and architecture.
Is cycle time the same as takt time?
Cycle time is not the same as takt time. Cycle time is the actual time a machine needs per product, while takt time is the pace that customer demand requires. A line only meets demand if its cycle time is shorter than the takt time.
📌 In summary
Cycle time is the duration of one complete PLC cycle of reading, computing and writing, and it determines how quickly and how predictably a control system responds. Choose the cycle time based on the fastest event in the process, assume two cycles as the worst-case response time, monitor minimum and maximum values and split the program across tasks. That keeps industrial automation stable and deterministic, and aligns the control system with the production cycle that counts in OEE.