Key Concepts in Time Domain Analysis

Why This Matters

Time domain analysis is the foundation of understanding how control systems actually behave. When you study control theory, you're being asked to predict what happens when someone flips a switch, changes a setpoint, or disturbs a system. Will the response be fast or sluggish? Will it overshoot dangerously or settle smoothly? Will there be persistent error?

These questions drive everything from autopilot design to industrial process control. The concepts here connect directly to system stability, transient performance, and steady-state accuracy. You'll see them appear in transfer function analysis, controller tuning problems, and system design questions. Don't just memorize definitions. Know what each parameter tells you about system behavior and how changing one affects the others.

---

System Characterization Methods

These foundational techniques reveal how a system behaves by examining its response to standardized inputs. Simple test signals (steps, impulses, ramps) expose complex system dynamics in predictable ways.

Step Response Analysis

The step response is the most common test signal in control systems. It evaluates how a system responds to a sudden, sustained change in input (like flipping a switch from 0 to 1). A single step test reveals stability, speed, and accuracy simultaneously, capturing rise time, settling time, overshoot, and steady-state error all at once.

This makes it the foundation for controller tuning. PID parameters are often adjusted based on step response characteristics because the step test is easy to perform experimentally and the resulting metrics map directly to design specifications.

Impulse Response Analysis

The impulse response $h(t)$ is mathematically fundamental because it completely characterizes any linear time-invariant (LTI) system. Taking the Laplace transform of $h(t)$ directly yields the transfer function $H(s)$.

The convolution relationship is what makes this so powerful: the output for any arbitrary input can be found by convolving that input with the impulse response. In practice, though, generating a true impulse is difficult, so step response testing is more common in the lab.

Ramp Response

The ramp response tests tracking ability by analyzing how well a system follows a continuously increasing input $r(t) = t$. This reveals velocity error: systems that perfectly track step inputs may still show persistent, growing error for ramp inputs.

Real-world relevance is high. Think of a radar dish tracking a moving aircraft or a motor maintaining speed on an incline. Both involve ramp-like reference signals.

---

Transient Response Parameters

These metrics quantify how a system gets to its final value. Transient behavior reflects the interplay between system speed and stability. Faster isn't always better if it causes oscillation.

Rise Time

Rise time measures response speed. It's typically defined as the time to go from 10% to 90% of the final value (though 0% to 100% is sometimes used, so check which convention your course follows).

• Inversely related to bandwidth: systems with higher bandwidth have shorter rise times
• Trade-off with overshoot: reducing rise time often increases overshoot unless damping is carefully controlled

Settling Time

Settling time defines practical stability: the time required for the response to stay within ±2% (or ±5%) of the final value permanently. For a standard second-order system:

$T_s \approx \frac{4}{\zeta \omega_n}$ (for the 2% criterion)

This formula shows that settling time is dominated by the product $\zeta \omega_n$, which equals the real part of the dominant complex poles. Many real applications specify a maximum allowable settling time as a hard design constraint.

Overshoot and Undershoot

Overshoot is quantified as a percentage of the final value:

$\%OS = \frac{\text{peak value} - \text{final value}}{\text{final value}} \times 100\%$

For a standard second-order system, overshoot depends only on the damping ratio:

$\%OS = e^{-\pi\zeta/\sqrt{1-\zeta^2}} \times 100\%$

Undershoot is a different phenomenon. It indicates non-minimum phase behavior, which occurs in systems with right-half-plane zeros. The initial response moves opposite to the final direction before correcting. This is physically meaningful: for example, an aircraft that briefly dips before climbing when you pull back on the stick.

Transient Response

"Transient response" is the umbrella term for all pre-steady-state behavior, encompassing rise time, overshoot, settling time, and oscillation characteristics. Two key principles:

• Pole locations determine the character of the response. Real poles produce exponential responses; complex conjugate poles produce oscillatory responses.
• Controllers are tuned to shape transient response. Most control design problems boil down to placing poles where they produce acceptable transient behavior.

---

Steady-State Performance

These concepts address what happens as $t \to \infty$. A system can have perfect transient response but still fail if it doesn't accurately track the desired output.

Steady-State Error

Steady-state error is the persistent difference between desired and actual output:

$e_{ss} = \lim_{t \to \infty} e(t)$

Two factors determine its value:

• System type (the number of free integrators in the open-loop transfer function). A Type 0 system has no integrators; Type 1 has one; Type 2 has two.
• Input type. Type 0 systems have finite error for step inputs. Type 1 systems have zero error for steps but finite error for ramps. Type 2 systems handle both steps and ramps with zero error.

Increasing loop gain reduces steady-state error, but higher gain can destabilize the system. This is one of the most fundamental trade-offs in control design.

Error Constants (Position, Velocity, Acceleration)

These constants give you a direct way to calculate steady-state error for unity feedback systems:

Each constant corresponds to a progressively more demanding input. A system with finite $K_p$ but $K_v = 0$ will track steps with some error but cannot track ramps at all (infinite error).

---

Time Constants and Delays

These parameters capture how fast a system responds and what lags exist in the response. Both fundamentally limit achievable closed-loop performance, but in very different ways.

Time Constants

For a first-order system $G(s) = \frac{1}{\tau s + 1}$, the time constant $\tau$ is the time for the response to reach 63.2% of its final value. It relates directly to pole location:

$\tau = \frac{1}{|p|}$

where $p$ is the pole. A pole at $s = -5$ gives $\tau = 0.2$ seconds.

In higher-order systems, each pole contributes its own time constant. The dominant time constant (from the slowest pole, closest to the imaginary axis) often determines overall response speed. This is why the dominant pole approximation works: faster poles decay quickly and contribute little to the long-term response.

Time Delay Effects

Pure time delay (also called dead time or transport lag) is modeled as $e^{-sT_d}$ in the transfer function. Unlike poles and zeros, a delay represents pure transportation lag with no dynamics.

Why delays are so problematic:

• They add phase lag without changing magnitude, which directly reduces phase margin
• They cannot be eliminated by feedback. You can't "cancel" a delay the way you can compensate for a pole
• They fundamentally limit achievable closed-loop bandwidth

---

Stability and System Structure

These analytical tools determine whether a system will behave predictably or diverge. Stability is non-negotiable: an unstable system is useless regardless of other performance metrics.

Stability Analysis in Time Domain

Three main approaches to stability analysis:

• BIBO stability (Bounded-Input, Bounded-Output): a system is BIBO stable if every bounded input produces a bounded output. This requires all closed-loop poles to lie in the left-half of the $s$-plane.
• Routh-Hurwitz criterion: an algebraic test that uses the coefficients of the characteristic polynomial to determine stability without solving for the actual pole locations. You construct a Routh array and check for sign changes in the first column; each sign change indicates one right-half-plane pole.
• Lyapunov methods: an energy-based approach primarily used for nonlinear systems. If you can find an energy-like function that continuously decreases along system trajectories, the system is stable.

Pole-Zero Analysis

Poles determine the natural response modes. Each pole contributes an exponential or oscillatory term to the time response:

• A real pole at $s = -a$ contributes a term $e^{-at}$
• Complex conjugate poles at $s = -\sigma \pm j\omega_d$ contribute a damped sinusoid $e^{-\sigma t}\sin(\omega_d t + \phi)$

Zeros affect response amplitude and shape. They can reduce overshoot (left-half-plane zeros near dominant poles) or cause non-minimum phase behavior (right-half-plane zeros).

The dominant poles approximation says that poles closest to the imaginary axis dominate the response because their associated terms decay most slowly. Poles far to the left decay quickly and can often be neglected for approximate analysis.

Root Locus Method

Root locus tracks how closed-loop pole locations move as feedback gain $K$ varies from 0 to $\infty$. It serves as both an analysis and design tool:

• Stability assessment: the system becomes unstable when the locus crosses into the right-half plane. The gain value at the crossing is the maximum stable gain.
• Gain selection: you choose $K$ to place dominant poles at locations that give the desired damping ratio and natural frequency.

---

Mathematical Representations

These frameworks provide the language for analyzing and designing control systems.

State-Space Representation

State-space describes a system using matrix equations:

$\dot{x} = Ax + Bu$ $y = Cx + Du$

where $x$ is the state vector, $u$ is the input, and $y$ is the output.

Why state-space matters:

• It handles MIMO (multiple-input, multiple-output) systems naturally, without the partial fraction complications of transfer functions
• It reveals internal dynamics that may be hidden in the transfer function, including unobservable and uncontrollable modes
• Modern control techniques (optimal control, state observers, Kalman filtering) all require state-space formulation

---

Quick Reference Table

---

Self-Check Questions

1. A second-order system has 15% overshoot and 2-second settling time. If you increase the damping ratio, what happens to both overshoot and rise time?

2. Which error constant ($K_p$, $K_v$, or $K_a$) would you use to find steady-state error for a system tracking a linearly increasing reference? What system type is required for zero error?

3. Compare the information provided by step response analysis versus impulse response analysis. Why might an engineer prefer one over the other in practice?

4. A system has poles at $s = -1$ and $s = -10$. Which pole dominates the transient response, and approximately what is the system's time constant?

5. Explain why time delay is more problematic for stability than a large time constant, even if both slow down the system response. How does each affect phase margin differently?