Engineering Notes · PLC Execution · 2026 Update
By KOEED Engineering · 2026-06-30 · 7 min read · Latest Blog
> AI Summary
- The PLC scan cycle still has three phases — input scan, program execution, output update — but in 2026 the timing budget is dominated by IEC 61131-3 fourth-edition semantics.
- Deterministic execution is no longer optional on safety-related machines: IEC 61131-3 Ed.4 and PL e / SIL 3 require a watched scan watchdog.
- Modular PLCs (Allen-Bradley ControlLogix, Siemens S7-1500, Mitsubishi iQ-R) now ship multi-tasking runtimes that split programs into FAST / NORMAL / SLOW tasks.
- Event-driven interrupts have replaced many legacy timed scans on packaging, motion, and vision lines.
- For KOEED customers running 1990s SLC 500 or S7-300 systems, the practical path is parallel stocking: keep legacy spares and migrate at the next outage.
The 2026 PLC scan cycle keeps input → execute → output, but IEC 61131-3 Ed.4 multi-tasking and watched safety tasks have reshaped determinism. What is new and what is unchanged.
If you commissioned a PLC in 2004, the scan cycle was a single loop: read inputs, run the program top to bottom, write outputs, repeat. In 2026 that mental model still works, but the runtime has grown layered tasks, event-driven I/O updates, and a safety layer that refuses to let the watchdog stretch. This Engineering Note walks through what changed, what stayed, and where engineers most often stumble.
The three-phase scan cycle — still the foundation
Every PLC, from a micro Allen-Bradley MicroLogix 1100 to a flagship Siemens SIMATIC S7-1500, executes a scan cycle. The cycle has three phases:
-
Input scan — read the physical state of every input module into the process image.
-
Program execution — execute the user program against the process image in the order defined by the runtime.
-
Output update — write the result image back to the physical output modules.
What 2026 runtimes add on top of this loop is the task scheduler. Where IEC 61131-3 third edition expected a single program scanned repeatedly, the fourth edition (and current vendor platforms) treat each program unit as a task with its own priority and interval.
> Tip
When you spec a replacement CPU, ask the vendor for the worst-case scan time, not the typical one. On a Siemens S7-1500 this is reported in the CPU properties; on ControlLogix it shows up in the task properties dialog. Use that number, not the marketing datasheet, when sizing your watchdog.
What changed since 2024: IEC 61131-3 Ed.4, multi-tasking, and watched safety
Three shifts matter for engineers planning new lines or planning migrations off legacy PLCs:
| Shift |
2022–2024 practice |
2026 practice |
| Cycle model |
Single scan, fixed watchdogs |
Prioritised multi-task runtimes |
| Safety integration |
External safety relay |
Watched safety task over CIP Safety / FSoE |
| I/O update |
At end of scan |
Event-driven for motion and vision |
| Diagnostics |
LED blink codes |
Time-stamped scan logs exported to MQTT |
1. Multi-tasking runtimes are the default
On Allen-Bradley ControlLogix you can now mix periodic tasks at 5 ms, 10 ms, 20 ms and 50 ms in the same project. Mitsubishi iQ-R and S7-1500 have similar split priority classes. The takeaway: scan time is no longer a single number — you have to budget the worst-case for each priority class.
2. Watched safety tasks (SIL 3 / PL e)
Safety PLCs from Schneider Modicon M580 Safety, Siemens F-CPU, and Allen-Bradley GuardLogix now expose a watched safety task. If the runtime exceeds the SIL 3 watchdog, the system falls to a safe state with no override. This is one of the bigger field stories of 2026 because legacy code assumptions ("the safety relay will catch it") no longer hold.
3. Event-driven I/O for motion and vision
For motion, the servo drive's cyclic position loop is no longer nested inside the PLC scan. EtherCAT distributed clocks and CIP Motion let the drive update its own outputs at sub-millisecond intervals while the PLC orchestrates the recipe. This is the practical reason you see Yaskawa Sigma-7 and Panasonic FP7 datasheets quoting "synchronised" cycles rather than "responding" cycles.
The execution model, phase by phase (2026)
Pulled together, the modern execution model on a modular PLC runs like this:
| Phase |
What runs |
Typical budget (2026 modular PLC) |
| 1. Input scan |
Backplane read into process image |
0.5–3 ms |
| 2. Priority task scheduler |
FAST → NORMAL → SLOW tasks |
2–20 ms aggregate |
| 3. Safety task (if enabled) |
Watched SIL 3 / PL e program |
1–5 ms, watched |
| 4. Output update and comms |
Process image → modules, EtherNet/IP, Profinet |
1–4 ms |
! Warning
Do not size your watchdog from the "typical" cycle you see in the IDE. Always measure the worst-case cycle under heavy I/O load and large tag-database operations. KOEED sees at least two support tickets a quarter where a safety stop triggered because the average cycle was 8 ms but peaks hit 28 ms.
What stayed the same since 2024 (and will likely stay in 2030)
A few rules from the 2024 article still hold and we are confident they will survive the next decade:
- Inputs are always sampled into the process image at the start of the cycle — never use a raw physical input inside a program branch.
- Outputs are always written at the end of the cycle — coil and assignment semantics are unchanged across ladder, FBD, ST, and SFC.
- Ladder logic remains the lingua franca for maintenance technicians because the wiring diagram still maps one-to-one onto the scan.
- PLCs continue to be the most reliable component in the cabinet — environmental tolerance, deterministic execution, and 20+ year service lives are still the design baseline.
Practical implications for KOEED customers
The lesson from a decade of cross-referencing scanned PLCs: do not migrate just to migrate. Keep legacy CPUs in your spares cabinet for the next planned outage, and use the maintenance window to move to a modern platform only when the spare-part risk is intolerable. A typical path we see for retrofits in 2026:
- Source a tested Allen-Bradley 1756-L73 or 1756-L75 for the existing ControlLogix chassis while it is still in active stock.
- Pre-order the replacement CPU (e.g. Siemens 6ES7 516 for S7-1500) and migration kit before the next planned shutdown.
- Document ladder logic into structured text only if the new project demands tag-based addressing.
- Move safety logic into the watched safety task — never leave it bolted onto the standard task.
Planning a scan-time or safety upgrade?
Send your BOM and migration BOM to Moritta@KOEED.COM. Active stock, EOL stock and cross-references — one quote within 24 hours.
Send My BOM →
Frequently asked questions
Does IEC 61131-3 Ed.4 change how ladder logic is scanned?
No. The scan still reads inputs, evaluates the program top-to-bottom, and writes outputs. Ed.4 adds a task scheduler and stricter type checking, but the ladder rung semantic is unchanged.
How do I size the watchdog on a modular PLC in 2026?
Multiply the slowest task's worst-case cycle by 2-3 to size the watchdog. SIL 3 applications must use the watched safety task, not a user timer; the runtime enforces a hard maximum.
Should I keep my legacy SLC 500 / S7-300 spares?
Yes. Both families are EOL but still widely running. KOEED stocks tested 1747, 6ES7 300 and Q-series modules for plants that need to bridge to their next planned outage.
What is the difference between a periodic and an event-driven task?
Periodic tasks run on a fixed interval (e.g. 10 ms). Event-driven tasks fire on a hardware trigger like an encoder zero or Profinet alarm. Use periodic for regulation; event-driven for I/O bursts.
Can multi-tasking runtimes be ordered by the same priority?
No. PLC runtimes enforce strict priority order. Same priorities are arbitrated by the scheduler but cause non-deterministic timing. Always assign distinct priorities to safety vs. standard tasks.
How do I get a quote for both legacy and current modules in one BOM?
Email your instrument list to Moritta@KOEED.COM and KOEED will return availability and lead time for both the legacy spare and the modern replacement, usually within 24 hours.
