Non-Linear Systems — Types of Non-Linearities in Control Systems
What is a Non-Linear System?
A non-linear system is any system that does not satisfy the principles of homogeneity (scaling) and superposition (additivity). In simple terms, if doubling the input does not double the output, or if the combined response to two inputs is not equal to the sum of individual responses, the system is non-linear.
In a linear system, the output waveform preserves the shape of the input — a sinusoidal input produces a sinusoidal output. In a non-linear system, a sinusoidal input can produce harmonics, sub-harmonics, or completely distorted output waveforms.
Homogeneity: f(αx) = α·f(x)
If either fails → System is Non-Linear
Most real-world physical systems are inherently non-linear. Linear models are approximations valid only within a limited operating range. When the system operates beyond this range, non-linear effects become dominant and must be accounted for in the design.
Linear vs Non-Linear Systems — Key Differences
Classification of Non-Linearities
Non-linearities in control systems are broadly classified into two categories:
1. Static Non-Linearity (Incidental)
The input-output relationship is described by a non-linear algebraic equation (no differential terms). The output depends only on the present value of input, not on its history. These are also called incidental non-linearities because they are undesirable but unavoidable in physical components.
2. Dynamic Non-Linearity (Intentional)
The input-output relationship is described by a non-linear differential equation. The output depends on both the present input and its rate of change or past values. Some dynamic non-linearities are intentionally introduced to improve system performance — for example, relay controllers in on-off systems.
Dynamic: dy/dt = f(x, y, t) — differential, has memory
Common Types of Non-Linearities
The following are the most frequently encountered non-linearities in control systems and electrical machines:
1. Saturation
In saturation, the output increases linearly with input up to a certain limit, beyond which it remains constant regardless of further input increase. This is the most common non-linearity in electrical systems.
Example: The B-H curve of an iron-cored coil. At low currents, inductance is linear. As current increases, the magnetic core saturates and inductance becomes non-linear. Operational amplifiers also exhibit output voltage saturation at ±Vsat.
 |
| Saturation characteristic — output clamps beyond ±M |
y = ±M, for |x| > a (saturated region)
Where K = slope, M = maximum output, a = saturation threshold
2. Dead Zone (Dead Band)
In a dead zone non-linearity, the output remains zero until the input exceeds a certain threshold value (±δ). Only after the input crosses this threshold does the output begin to respond. This is common in mechanical systems with friction and in electronic amplifiers with crossover distortion.
Examples: DC servomotors (coulomb friction prevents rotation for small voltages), gear trains with play, and Class-B push-pull amplifiers.
 |
| Dead zone — no output for |x| < δ |
y = K(x − δ), for x > δ
y = K(x + δ), for x < −δ
3. Backlash (Hysteresis in Mechanical Systems)
Backlash occurs in gear trains and mechanical linkages where a small gap exists between mating parts. When the input reverses direction, the output does not immediately follow — it remains stationary until the gap is taken up. This introduces a phase lag and can cause limit cycle oscillations in closed-loop systems.
4. Relay (On-Off Non-Linearity)
A relay produces only two output levels — fully ON or fully OFF — depending on the sign of the input. This is an intentional non-linearity used in bang-bang controllers, thermostats, and spacecraft attitude control. Variants include relay with dead zone and relay with hysteresis.
Relay with Dead Zone: y = 0 for |x| < δ
Relay with Hysteresis: switching thresholds differ for increasing vs decreasing input
5. Friction (Coulomb + Viscous)
Real mechanical systems exhibit multiple friction types simultaneously. Coulomb friction is constant and opposes motion direction. Viscous friction is proportional to velocity. Stiction (static friction) is higher than coulomb friction and must be overcome to initiate motion. The combination creates a non-linear torque-speed characteristic that causes stick-slip oscillations.
Analysis Methods for Non-Linear Systems
Since transfer function and Laplace transform methods are not applicable to non-linear systems, special techniques are used:
Real-World Applications
Understanding non-linear systems is critical in:
- Power Electronics: Switching converters (PWM inverters, SMPS) are inherently non-linear due to on-off switching
- Electric Machines: Magnetic saturation in transformers and motors affects performance at high flux densities
- Robotics: Joint friction, backlash in gears, and actuator saturation limit precision
- Aerospace: Relay-based attitude control (bang-bang controllers) for spacecraft
- Power Systems: Transformer core saturation causes harmonic distortion in the grid
Frequently Asked Questions
Q1: Why can't we use transfer functions for non-linear systems?
Transfer functions are defined using the Laplace transform, which requires the superposition principle. Since non-linear systems violate superposition, the transfer function concept is not applicable. Instead, we use describing functions, phase plane, or Lyapunov methods.
Q2: What is the difference between static and dynamic non-linearity?
Static non-linearity has no memory — the output depends only on the current input value (algebraic relationship). Dynamic non-linearity involves differential equations where output depends on input history and rate of change.
Q3: Can a non-linear system be stable?
Yes, but stability depends on the input magnitude and initial conditions, unlike linear systems where stability is a system property. A non-linear system may be stable for small perturbations but unstable for large ones (conditional stability).
Q4: What is a limit cycle in non-linear systems?
A limit cycle is a sustained oscillation of fixed amplitude and frequency that occurs in non-linear systems without any external periodic input. It is a closed trajectory in the phase plane. Relay systems and systems with backlash commonly exhibit limit cycles.
Q5: What is the describing function method?
The describing function is a quasi-linear approximation that represents a non-linear element by its fundamental harmonic response to a sinusoidal input. It extends the Nyquist criterion to predict whether limit cycles exist and their approximate amplitude and frequency.