Controller

What is a controller?

A controller is a device or software component that automatically operates a process, machine or installation by reading measurements from sensors, comparing them with a desired value or logic, and driving actuators such as valves, motors and pumps accordingly. Controllers are the heart of industrial automation: they make the decisions that keep a plant running, every millisecond to second. Think of a PLC coordinating a packaging line or a loop controller holding an oven at temperature. The best-known form is the PLC, but a DCS, an RTU, a motion controller and an embedded controller are all controllers too. In the Purdue Model they sit at level 1, directly above the field devices.

🧠 What exactly does a controller do?

Every controller runs through a fixed cycle, regardless of brand or type:

• Read - the controller reads digital and analogue inputs, for example a limit switch or a 4-20 mA signal from a pressure transmitter
• Compare - the measured value is compared with a setpoint or with conditions in the program (“tank full and pump running?“)
• Decide - logic, a control algorithm such as PID or a computational model determines what should happen
• Act - the controller sets its outputs: a valve opens, a variable frequency drive receives a new speed
• Feed back - the next measurement shows whether the action had the intended effect; this forms a closed control loop or feedback loop

Compared with manual operation or hard-wired relay logic, this delivers greater precision, speed and repeatability, and industrial controllers are built to run reliably for years despite vibration, dust and temperature swings. Operators still supervise and step in when faults or deviations occur.

In a PLC this cycle is called the scan. A typical scan time is roughly 1 to 100 ms, depending on program size and processor. Tasks that need a fixed rhythm, such as a control loop, are often placed in a periodic task with a fixed interval.

🔧 Which types of controllers exist?

The term PAC (Programmable Automation Controller) was introduced in the early 2000s by ARC Advisory Group to distinguish controllers that combine logic, motion, drives and process control with a single database. In practice the difference from a modern high-end PLC has become largely a matter of marketing.

The cycle times listed are indicative: the actual value depends on hardware, program and configuration.

🕰️ How did controllers come about?

Before the controller, control meant relay cabinets and pneumatic or electronic loop controllers. Every change meant rewiring. In 1968 the Hydra-Matic division of General Motors asked for an electronic replacement for relay logic; in 1969 Bedford Associates delivered the Modicon 084, regarded as the first PLC (see Modicon). The mathematical basis of PID control is older: Nicolas Minorsky analysed automatic ship steering in 1922, and Ziegler and Nichols published their well-known tuning method in 1942.

Since then controllers have increasingly become computers. The IEC 61131-3 standard has defined the programming languages since 1993; edition 4 appeared in May 2025 and covers Structured Text, Ladder Diagram, Function Block Diagram and Sequential Function Chart, while Instruction List has been dropped. In addition, IEC 61499 describes a model for distributed control with function blocks, and initiatives such as Open Process Automation aim for vendor-independent controllers.

🔄 What is the difference between a controller, SCADA and an HMI?

A controller performs the control itself, deterministically and in real time. SCADA and the HMI supervise, show operators the state of the process and let them change setpoints or start equipment; the controller then executes that command. MES systems at level 3 manage production orders but do not act directly on valves or motors.

This distinction matters for design: critical control loops belong in the controller, not in a SCADA script or in the cloud.

🏭 How do you choose and design a controller in practice?

When selecting and designing a controller, experienced engineers work through a fixed set of questions:

1. Determine the process type - discrete manufacturing (on/off, sequences) points to a PLC; a continuous process with many control loops to a DCS; dispersed sites to RTUs
2. Establish timing requirements - what response time is needed? An oven temperature loop tolerates hundreds of milliseconds, a servo axis does not
3. Inventory the I/O - number of digital and analogue signals plus spare capacity (often 10–20%), local or O
4. Separate safety - safety functions get their own safety controller according to IEC 61508 or IEC 61511 (process) or ISO 13849-1 and IEC 62061 (machinery), independent of the process control
5. Arrange availability - critical processes get redundant CPUs and power supplies
6. Choose communication - PROFINET, IP, EtherCAT or Modbus TCP, in line with the rest of the installation
7. Plan the life cycle - controllers often run for 15–25 years; watch spare-part availability and firmware support (lifecycle management)

Example: a Dutch drinking water company controls a pumping station with a PLC that keeps the level in a clean-water reservoir at setpoint using a PID loop acting on the pumps’ frequency drives. An RTU or the PLC itself sends measurements and alarms to the central SCADA. If the connection drops, the local controller keeps running the station safely. The same principle applies in an office building, where an HVAC controller regulates ventilation, heating and cooling against temperature and CO₂ setpoints.

🔐 Why are controllers a target for attackers?

Controllers are designed for reliability, not for security. Many older protocols have no authentication: anyone who reaches the network can read or overwrite the logic. Well-known incidents show what that means:

• Stuxnet reprogrammed Siemens PLCs at the Iranian enrichment plant in Natanz and hid the changes from the operators
• TRITON targeted a Triconex safety controller at a petrochemical plant in 2017
• CyberAv3ngers took over internet-exposed Unitronics PLCs in late 2023, including at a drinking water utility, where the default password “1111” was still active

Key measures are network segmentation according to IEC 62443, no direct internet connection, changing default passwords, the physical key switch in RUN mode, signed firmware, backups of the program and change management via a secured engineering station. For vendors, IEC 62443-4-2 sets security requirements for components such as embedded devices. European rules raise the bar further: the Cyber Resilience Act introduces reporting obligations from 11 September 2026 and applies in full from 11 December 2027, and the Machinery Regulation requires protection of control systems against corruption from 20 January 2027.

❓ Frequently asked questions

Is a PID controller the same as a controller?

A PID controller is a control algorithm that keeps a process variable at setpoint; it is a function that runs on a controller. A PLC or DCS controller can execute hundreds of PID loops at the same time. A compact process controller is a small controller that mainly runs PID loops.

What is the difference between a PLC and a PAC?

A PAC is a controller with more computing power, memory and communication options that combines logic, motion and process control on one platform. The term was introduced in the early 2000s by ARC Advisory Group. Modern PLCs now offer similar capabilities, so the distinction mostly survives in marketing.

How fast does an industrial controller respond?

A PLC typically runs through its program in about 1 to 100 ms, a motion controller often synchronises axes within 1 ms, and a DCS usually executes control loops every 100 ms to 1 s. The required speed depends on the process. A thermal process is slow, whereas a servo axis needs a very short, fixed cycle time.

Can a safety function run in an ordinary PLC?

A safety function belongs in a certified safety controller developed according to IEC 61508 that achieves the required SIL (process industry, IEC 61511) or Performance Level (machinery, ISO 13849-1). A standard PLC lacks the required diagnostics and failure-probability calculation. Separating process and safety control also prevents a fault or attack on one layer from disabling the other.

Can a controller run in the cloud or as software?

A controller can run as a soft PLC or virtualised controller on an industrial PC or server, provided the platform guarantees deterministic behaviour. The cloud is unsuitable for real-time control because of latency and dependence on the connection. Cloud platforms are therefore used for analysis and optimisation, while the controller stays local.

How long does an industrial controller last?

An industrial controller often runs for 15 to 25 years in practice, far longer than an office PC. As a result, many installations still contain active controllers without current security updates. A migration plan and lifecycle management prevent a controller from becoming irreplaceable.

📌 In summary

A controller reads measurements, compares them with a setpoint or logic and drives actuators in a continuous cycle. PLCs, PACs, DCS controllers, RTUs, motion controllers and safety controllers differ in speed, scale and application, but share the same basic function. Because controllers act directly on the physical process, they deserve the strongest protection in any OT convergence effort.

A controller is the deciding brain at level 1 of every automated installation: choose it by process type and timing requirements, separate safety from control, and protect it as the most critical part of your OT.