Types of Controllers and Controller Tuning (MICROL, 2004)
A 63-page industrial note explaining open- and closed-loop regulation, dynamic response, two- and three-position controllers, pulse outputs, P/PI/PD/PID structures, and empirical tuning. Its block diagrams are useful, but controller-selection tables and tuning constants are heuristics rather than portable design guarantees.
MICROL's note is a compact industrial-control primer. It starts from a measurement–controller–actuator–plant loop, separates regulation by error from disturbance compensation, then introduces step and frequency response, common process elements, switching controllers, continuous P/PI/PD/PID control, pulse-width output, and several manual tuning rules.
Its strongest contribution is concrete actuator semantics. A “controller type” is not only a formula: the output may be a continuous analogue value, two logical states, two opposing outputs plus neutral, or timed pulses whose duty cycle drives a contactor, valve motor, heater, or fan. That distinction now lives in Relay, Hysteresis & Pulse Control, while the continuous law and anti-windup mechanisms remain in PID Control, Tuning & Anti-Windup.
Information-bearing visual audit #
All 63 pages were rendered and reviewed; 23 diagram-heavy pages were inspected again at readable resolution. The useful visual families are:
- open-loop, feedback and disturbance-feedforward block diagrams;
- step, impulse and sinusoidal test inputs with the corresponding process response;
- ideal and real proportional, integrating, differentiating, lag and oscillatory elements;
- hysteresis loops for two- and three-position controllers;
- pulse-output timing and multi-position switching;
- analogue, pulse and PWM actuator interfaces;
- P, PI, PD and PID step-response sketches.
No original visual is copied. The notation is tied to one vendor's 2004 product conventions, several diagrams collapse sensor and actuator dynamics, and the output-stage examples predate modern implementation concerns. The topic entries use independent diagrams that make sampling, saturation, switching wear, noise and safety boundaries explicit.
What remains technically useful #
- A switching controller needs a hysteresis band or another dwell mechanism; otherwise measurement noise and delay produce chatter and contact wear.
- A three-position output is genuinely different from a two-position output when it can command increase, hold and decrease, such as a motorized valve.
- Pulse and PWM outputs separate average requested effort from instantaneous actuator state; the plant must be slow enough, and the actuator rated, for that approximation to hold.
- Step-response features — gain, apparent delay, time constant, overshoot and settling — are diagnostics, not a complete plant model.
- P, I and D terms address different error histories and create different noise, saturation and robustness costs.
Source claims audit #
| Load-bearing source claim | Status | Treatment in the bank |
|---|---|---|
| negative feedback compares a reference with a measured output and acts on the error | current | retained with explicit sign, disturbance, sensor and actuator paths |
| process response can be characterized by step and frequency tests | current with limits | retained, but excitation, operating point, uncertainty and validation data are required |
| relay hysteresis reduces switching chatter | current | retained and quantified as a deliberate accuracy/wear trade-off |
| time-constant and delay ratios select a controller type universally | needs correction | treated as a screening heuristic; uncertainty, nonlinearity, constraints and robustness decide the design |
| a fixed table of P/PI/PID constants gives optimal tuning | error in source if read generally | presented only as an empirical starting point tied to a chosen test and performance criterion |
| derivative action can be used as an ideal differentiator | needs correction | every practical derivative path is filtered and usually acts on measurement, not setpoint |
| a digital controller is preferable merely because delay is large | error in source | sampling and computation do not remove plant delay; prediction or an appropriate dead-time design may be needed |
How to cite it #
Use this source for the vocabulary and physical output forms of early industrial controllers. Do not cite its selection table or tuning constants as design authority. Current explanations in the bank pair its practical examples with the control equations, robustness limits and implementation safeguards that the note omits.