Control Theory Mathematics: Laplace and Feedback

What Is Control Theory?

Control theory is the mathematical framework for designing systems that automatically maintain desired values despite disturbances. Every control system in a factory — temperature regulators, motor speed controllers, pressure regulators — is built on this mathematics.

The fundamental control problem: given a system (plant) affected by disturbances, design a controller that keeps the output at the desired reference value with acceptable speed, accuracy, and stability.

Laplace Transform: From Time to Frequency

The Laplace transform is the foundational mathematical tool in control theory. It converts differential equations (difficult to solve) into algebraic equations (easy to solve).

Definition:

F(s) = L{f(t)} = integral_0^inf f(t) * e^(-st) dt

Where s = sigma + j*omega is a complex variable.

Essential transform pairs:

The real power — transforming derivatives:

L{df/dt} = s*F(s) - f(0)
L{d^2f/dt^2} = s^2*F(s) - s*f(0) - f'(0)

Differentiation becomes multiplication by s — a differential equation becomes an ordinary algebraic equation.

Example: Spring-mass-damper equation:

m * d^2x/dt^2 + c * dx/dt + k * x = F(t)

After Laplace transform (with zero initial conditions):

(m*s^2 + c*s + k) * X(s) = F(s)

Transfer Functions: The System's Fingerprint

The transfer function is the ratio of the system output to its input in the Laplace domain:

G(s) = Y(s) / U(s)

Where Y(s) is the output and U(s) is the input. The transfer function completely describes the system's input-output behavior.

Common systems:

First-order system (thermal tank, RC circuit):

G(s) = K / (tau*s + 1)

Where K = static gain (final response value) and tau = time constant (response speed). After 5*tau, the response reaches 99% of its final value.

Second-order system (spring-mass-damper, motor with load):

G(s) = wn^2 / (s^2 + 2*zeta*wn*s + wn^2)

Where:

• wn = natural frequency (rad/s) — response speed
• zeta = damping ratio (dimensionless) — determines behavior:

Practical optimum: zeta = 0.7 — balances speed and minimal overshoot (less than 5% overshoot).

Block Diagrams: The Control Engineer's Language

Block diagrams represent systems graphically. Each block represents a transfer function, and arrows represent signals.

Basic closed-loop control:

  R(s) --> [+] --> [G_c(s)] --> [G_p(s)] --> Y(s)
            ^ -                               |
            |                                 |
            +-------- [H(s)] <----------------+

Where:

• R(s) = reference signal (setpoint)
• G_c(s) = controller transfer function (e.g., PID)
• G_p(s) = plant transfer function (motor, furnace, process)
• H(s) = sensor transfer function (temperature sensor, encoder)
• Y(s) = actual output

Closed-loop transfer function:

T(s) = G_c(s) * G_p(s) / (1 + G_c(s) * G_p(s) * H(s))

Block diagram reduction rules:

• Series: G1(s) * G2(s) — multiply transfer functions
• Parallel: G1(s) + G2(s) — add transfer functions
• Feedback: Use the formula above

Bode Plot: Frequency Domain Insight

The Bode plot shows a system's response across different frequencies. It consists of two graphs:

1. Magnitude plot: 20*log10|G(jw)| in dB versus frequency
2. Phase plot: angle of G(jw) in degrees versus frequency

For a first-order system G(s) = K/(tau*s+1):

• At low frequencies (w << 1/tau): Magnitude = 20*log(K) dB, Phase = 0 degrees
• At corner frequency (w = 1/tau): Magnitude drops by 3 dB, Phase = -45 degrees
• At high frequencies (w >> 1/tau): Magnitude rolls off at -20 dB/decade, Phase approaches -90 degrees

Reading a Bode plot — critical information:

• Gain margin: How much gain can be added before the system becomes unstable — measured at the frequency where Phase = -180 degrees
• Phase margin: How far the phase is from -180 degrees at the frequency where Magnitude = 0 dB
• Rule of thumb: Gain margin > 6 dB and Phase margin > 30 degrees ensure comfortable stability

Nyquist Plot and Root Locus

The Nyquist plot traces G(jw) in the complex plane (real part versus imaginary part) as frequency varies from 0 to infinity.

Nyquist stability criterion: The closed-loop system is stable if and only if the number of encirclements of the point (-1, 0) by the Nyquist contour equals the number of open-loop poles in the right half-plane.

Simply put: If the open-loop system is stable (no right half-plane poles), the closed-loop system is stable if the Nyquist curve does not encircle the point (-1, 0).

Root locus:

Plots the path of closed-loop poles as the controller gain K varies from 0 to infinity.

Key rules:

• Starts at open-loop poles (at K=0)
• Ends at open-loop zeros (at K approaching infinity)
• The system is stable as long as all poles remain in the left half-plane (Re(s) < 0)
• When a pole crosses the imaginary axis — the system loses stability

Stability Criteria

Routh-Hurwitz Criterion

An algebraic method to determine stability without actually computing the roots. From the characteristic equation:

s^4 + a3*s^3 + a2*s^2 + a1*s + a0 = 0

Build the Routh table — if all elements of the first column are positive, the system is stable. Any sign change indicates a pole in the right half-plane (instability).

Necessary condition (but not sufficient): All coefficients of the characteristic equation must be positive and present. If any coefficient is negative or zero — the system is definitely unstable.

PID Control Mathematics

The PID controller is the most widely used in industry (over 95% of control loops). Its transfer function:

G_c(s) = K_p + K_i/s + K_d*s

Or in standard form:

G_c(s) = K_p * (1 + 1/(T_i*s) + T_d*s)

Where:

• K_p = proportional gain — responds to the current error
• K_i = K_p/T_i = integral gain — eliminates steady-state error by accumulation
• K_d = K_p*T_d = derivative gain — anticipates future error

Effect of each term:

PID Tuning Methods

Ziegler-Nichols Method

The most well-known empirical tuning method — requires no mathematical model, only experimentation on the real system.

Method 1 — Step response:

1. Set the controller to manual mode
2. Apply a step change to the input
3. From the response curve, measure the delay time L and time constant T

Method 2 — Ultimate gain (critical oscillation):

1. Start with proportional control only (K_i = K_d = 0)
2. Gradually increase K_p until the system sustains continuous oscillation
3. Record the ultimate gain K_u and ultimate period T_u

Important note: Ziegler-Nichols provides a good starting point but requires manual fine-tuning — the resulting overshoot is typically 25%, which is excessive for many applications. Modern plants use more advanced techniques such as Optimal Control and Model Predictive Control (MPC).

From Theory to the Factory Floor

In industrial practice, most control systems run PID algorithms embedded in PLCs. Understanding the mathematics is not academic luxury — it is essential for:

• Troubleshooting: Why is the output oscillating? Is the problem with gain or damping?
• Performance tuning: Reducing settling time and minimizing overshoot
• Controller selection: When is P sufficient? When do you need PI? When is full PID required?
• New system design: Selecting sensors and actuators based on mathematical performance requirements