PID Controller Explained: Basics, Block Diagram, Transfer Function

Transient response analysis provides essential information about system dynamics including when the system begins responding, the rate of output rise indicating system speed, whether output increases exponentially or oscillates, overshooting behavior, and settling time. Steady-state response analysis reveals how close the actual output is to the desired value, indicating system accuracy, and whether errors remain constant or vary with time. Together, they provide comprehensive performance assessment. The total time response is expressed as C(t) = C_ss + C_tt, where C_ss is steady-state output and C_tt is transient output. The difference between desired and actual output is called steady-state error, which is crucial for system design.

This section covers two critical aspects of system performance: settling time and steady-state error. Settling time (ts) is defined as the time required for the response to remain within a specified tolerance band (typically 5% or 2%) around the final value. For practical purposes, ts ≈ 3/(ζωn) for 5% settling time and ts ≈ 4/(ζωn) for 2% settling time. Steady-state error analysis uses the final value theorem: ess = lim(s→0)[sR(s)/(1 + G(s))]. For step input, ess = 0 for all ζ > 0. However, for the undamped case (ζ=0), the system is marginally stable with poles on the imaginary axis, causing the output to oscillate indefinitely between values, resulting in an error range of -1 to +1 rather than a specific value.

System response consists of two parts: transient response (which decays over time) and steady-state response (which remains constant). The instructor explains that transient response is the part of system behavior that changes and eventually dies out, while steady-state response is what remains after the transient has decayed. Key time-domain parameters include rise time (time to go from 10% to 90% of final value), peak time (time to reach maximum value), and maximum overshoot (amount by which response exceeds final value). These parameters help characterize system performance and are essential for control system design.

Four key specifications quantify system transient response: (1) Settling time (TS) - time for response to enter and stay within ±2% of steady-state value; (2) Peak time (TP) - time to reach maximum overshoot; (3) Rise time (TR) - time from 10% to 90% of steady-state value; (4) Percentage overshoot (MP%) - [(max response - steady-state)/steady-state] × 100. These metrics enable precise specification of desired system behavior for control design. The settling time definition assumes asymptotic stability, meaning the system must reach a finite steady-state value.

Transient response analysis quantifies system behavior using measurable metrics including overshoot, settling time, and peak time. These metrics balance speed against overshoot constraints in applications like satellite control and robot grippers. First order systems with feedback have transfer function Y(s)/R(s) = k/(τs + k), where time constant τ = τ_plant/k. Increasing control gain k decreases the time constant, making the system respond faster. Second order systems have transfer function Y(s)/R(s) = k/(s² + as + k), converting to standard form ωn²/(s² + 2ζωns + ωn²) with ωn = √k and ζ = a/(2√k). Unlike first order systems, increasing k decreases ζ (making the system more underdamped) while increasing ωn. As k increases, poles move from the real axis toward the imaginary axis, changing behavior from overdamped to underdamped. Five key performance metrics characterize transient response: rise time, peak time, percent overshoot, settling time, and steady-state value. The percent overshoot formula PO = 100 × exp(-πζ/√(1-ζ²)) is ONLY valid for underdamped systems (0 < ζ < 1).

This comprehensive section covers frequency domain stability analysis methods. Vector locus traces G(jω) as ω varies from 0 to infinity. Gain margin (GM) is calculated as 20 log₁₀(1/|G(jω)H(jω)|) at ω=0, representing decibel gain increase before instability. Nyquist criterion states GM > 0 dB implies stability. Bode plots use gain and phase curves with gain crossover (ωc) and phase crossover (ωp) frequencies. Phase margin (φm) and gain margin quantify stability margins. Time domain methods (Routh) determine stability but cannot assess improvement potential, while frequency domain methods provide quantitative margins. Root locus traces characteristic equation roots as gain K varies from 0 to infinity. Zeros make numerator zero; poles make denominator zero. The number of root locus branches equals the maximum of zeros and poles, or the highest characteristic equation degree.

This extensive section covers frequency-domain stability analysis using Nyquist and Bode plots. The Nyquist criterion determines stability by examining whether the open-loop frequency response trajectory encircles the critical point (-1, j0). A system is stable if the plot does not encircle this point. Gain margin (GM) and phase margin (φm) quantify relative stability: GM measures how much gain can increase before instability occurs, while φm measures additional phase lag tolerance. Bode plots provide an alternative approach using separate magnitude and phase versus frequency plots. The system is stable if, at the gain crossover frequency (0 dB), phase exceeds -180°, and at the phase crossover frequency (-180°), magnitude is below 0 dB. Both methods enable quantitative assessment of system robustness against parameter variations.

Nyquist plots can analyze systems with unstable open loop plants, whereas Bode plots may give incorrect results for such systems. Nyquist provides both stability and stability margins, while Bode is better suited for frequency response analysis. Use Nyquist for stability determination and Bode for frequency analysis when computer tools are available.

Frequency domain analysis involves studying system behavior using Nyquist and Bode plots. Nyquist plots show the polar plot of the open-loop transfer function in the complex plane. Bode plots show the magnitude and phase of the transfer function versus frequency. These plots are used for stability analysis and controller design. The Nyquist stability criterion determines system stability by analyzing the encirclement of the critical point (-1, j0) in the Nyquist plot.

Nyquist stability criterion: The number of clockwise encirclements of (-1, j0) equals the number of unstable poles. For stable closed-loop systems, unstable poles must be zero. Bode plot analysis: Find frequency where phase shift equals -180°, then check amplitude ratio. If amplitude ratio < 1, the system is stable. The instructor emphasizes that students should understand these methods as they are frequently tested in GATE exams.

Two phenomena commonly occur in PID controllers: (1) Derivative Kick: When a step change in reference occurs, the derivative term responds to the rapid change in error, causing an initial spike in the control output. This occurs because the derivative term is proportional to the rate of change of error. (2) Integral Windup: When the system output exceeds the reference, the error becomes negative, and the integral term accumulates this error. If the actuator cannot apply negative control (due to saturation), the integral term continues to accumulate, causing the controller to 'wind up.' This results in the controller applying excessive corrective action even after the error has changed sign, causing overshoot and slow response.

Integral windup occurs when the integral term continues accumulating error even when the actuator has saturated. In the drone example, when held down, the constant altitude error causes the integrator to request propeller speeds beyond the motor's maximum capability. When released, the system overshoots significantly because the integral must first decrease from its accumulated value before the actuator can respond. Anti-windup methods prevent this by clamping the integrated value to prevent it from exceeding a specified limit, ensuring immediate response when error changes sign.

Wind-up is a phenomenon where the integral term accumulates excessive error values beyond the physical limits of the control system. Mathematically, the integral can grow indefinitely, but physically, actuators have saturation limits (maximum voltage, current, force, etc.). When the integral term exceeds these limits, the controller output saturates, causing the system to respond slowly and potentially become unstable. Anti-windup compensation prevents this by clamping the integral term to the saturation limits when the controller output reaches its maximum or minimum value. The derivative term can also benefit from anti-windup protection by clamping it to its saturation limit. This prevents the derivative action from responding excessively to noise or rapid changes in the error signal, which could cause instability or excessive control effort.

Derivative control predicts future trends by measuring error rate-of-change, anticipating overshoot before it happens. When systems saturate at maximum power, error accumulates causing oscillatory behavior known as windup. Anti-windup mechanisms detect saturation and bypass the integrator to prevent further error accumulation. Complete PID circuits include proportional, integral, and derivative stages with independent adjustment pots, plus anti-windup protection using zener diodes and noise filtering for derivative inputs.

Integral windup occurs when the controller output reaches its physical limit (saturation) while the integral term continues accumulating error. Using the car analogy, if a small engine cannot climb a steep hill at 60 mph, the controller keeps pushing the pedal to the floor (maximum output) while the integral term accumulates positive error. When the hill ends and the car goes downhill, the accumulated integral term causes the controller to demand excessive acceleration (trying to compensate for the accumulated positive error), potentially causing dangerous overspeed conditions. Anti-windup strategies prevent this by stopping integration when the controller output is saturated.

Cascade control systems use multiple nested feedback loops where an inner loop handles fast-acting disturbances (like motor speed) while an outer loop manages slower processes (like drone altitude), enabling better disturbance rejection and allowing different teams to work on separate components; discrete PID controllers, which run on digital computers at fixed time intervals, require careful consideration of sampling rates to maintain performance similar to continuous-time controllers, with typical design approaches involving tuning in the continuous domain first and then converting to discrete form.

Advanced control strategies extend beyond single-loop PID control. Cascade control uses nested loops where a primary controller sets the set point for a secondary controller, providing faster disturbance rejection for processes with significant lag. Flow control serves as the secondary loop in temperature-flow cascades. PID control offers advantages over PI control in processes with substantial lag through phase cancellation between integral and derivative actions. However, significant process dead time limits PID effectiveness—when delay exceeds the process time constant, PID control fails. These advanced strategies address limitations of basic PID control in complex industrial processes requiring tighter control and faster response times.

Advanced control strategies include cascade control with master-slave architectures, feedforward control for disturbance anticipation, and complex algorithm combinations. Cascade control uses multiple controllers where a master controller sets setpoints for slave controllers, improving response to disturbances. Feedforward control provides anticipatory action by detecting disturbances before they affect controlled variables, improving response time. Control algorithms combine proportional, integral, and derivative actions (PID) based on process characteristics. Final control elements include modulating valves with electric or pneumatic actuators, on-off valves, frequency converters for energy savings, and servo mechanisms. Control system design requires careful consideration of process dynamics, tuning requirements, and safety constraints.

This section covers advanced control strategies and their industrial implementation. Topics include: (1) Systematic control structure design using cascade, feedforward, override, ratio, split-range, and mid-range control strategies; (2) PID controller fundamentals and tuning methods; (3) PID controller recycling for complex processes with dead time and nonlinearities; (4) Internal Model Controller (IMC) construction and application; (5) Controller operating modes: automatic/manual, external/internal setpoint, and output tracking/interlock modes; (6) Bumpless transfer requirements for smooth mode transitions; (7) Alarm handling with warning/alarm limits, acknowledgment, and history logging. The section emphasizes applying theoretical knowledge to real industrial systems through practical examples and simulation.

This lecture introduces three advanced control strategies—cascade, ratio, and feed forward control—to improve feedback controller performance by eliminating specific disturbances. The fundamental principle is that corrective action occurs by measuring disturbances before the control variable changes significantly. These methods differ in implementation: cascade and ratio are incorporated within feedback loops, while generic feed forward operates as an independent controller. Effective application requires identifying which disturbance type matches each strategy's requirements.