Stability Analysis from Pole Locations

Understanding system stability—whether a system's output will remain bounded over time—is arguably the most critical first step in control systems engineering. For Linear Time-Invariant (LTI) systems, this abstract concept is powerfully and elegantly visualized on the s-plane, where the locations of a system's poles serve as a direct map to its dynamic behavior.

The Foundational Link: Poles and Impulse Response

To grasp why pole locations dictate stability, you must first recall what a pole represents. In the Laplace domain, the transfer function of an LTI system is a ratio of polynomials in the complex variable $s$. The poles are the values of $s$ that make the denominator polynomial zero. Each distinct pole $s=p$ corresponds to a specific mode in the system's natural response—the behavior you see when you tap the system (like an impulse input) and let it react.

The connection to time is given by the inverse Laplace transform. A real pole at $s=p$ corresponds to an exponential time function $e^{pt}$. A complex conjugate pole pair at $s=\sigma \pm j\omega$ corresponds to a sinusoidal mode modulated by an exponential: $e^{\sigma t}\sin (\omega t+\varphi )$. The real part of the pole, $\sigma$, is the exponential's growth/decay rate. This is the key: the sign of the real part of every pole determines the fate of its corresponding mode as time progresses.

Interpreting the S-Plane: Left-Half, Right-Half, and the Imaginary Axis

The s-plane is a complex coordinate system where the horizontal axis represents the real part (Re(s) or $\sigma$) and the vertical axis represents the imaginary part (Im(s) or $\omega$). Stability is assessed by where poles lie relative to the imaginary axis (the vertical line at $\sigma =0$).

1. Left-Half s-Plane (LHP): A system is asymptotically stable if and only if all of its poles have negative real parts ($\sigma <0$). Why? Each mode takes the form $e^{\sigma t}$, and if $\sigma$ is negative, this exponential decays to zero over time. Any oscillations ($\sin (\omega t)$) associated with complex poles are also damped out. The system's natural response eventually vanishes, returning to equilibrium.
2. Right-Half s-Plane (RHP): If any pole has a positive real part ($\sigma >0$), the system is unstable. The corresponding mode $e^{\sigma t}$ grows without bound. Even if other poles are stable, this single growing mode will dominate the system's output, leading to runaway behavior.
3. Imaginary Axis: Poles that are purely imaginary ($\sigma =0,s=\pm j\omega$) create a special condition called marginal stability (or "limited stability"). Here, the exponential factor $e^{0t}=1$, leaving a persistent, undamped oscillation like $\sin (\omega t)$. The output remains bounded but does not decay to zero. Crucially, for true marginal stability, these imaginary-axis poles must be simple (not repeated). A repeated pole on the imaginary axis introduces a mode that grows polynomially (e.g., $t\sin (\omega t)$), leading to instability.

The Routh-Hurwitz Criterion: Stability Without Explicit Poles

Finding the exact pole locations for high-order systems by solving the characteristic polynomial can be algebraically tedious or numerically intensive. The Routh-Hurwitz criterion is an algebraic test that determines the number of RHP poles without factoring the polynomial. You apply it to the system's characteristic equation, which is the denominator of the transfer function set equal to zero: $$a_n{s}^n+a_{n-1}{s}^{n-1}+...+a_{1}s+a_0=0$$

Step 1: Form the Routh Array. The first two rows are filled with the polynomial coefficients. Subsequent rows are calculated using a determinant-like pattern.

Where: $$b_1=\frac{-\left|\begin{array}{cc}{a}_n& a_{n-2}\\ a_{n-1}& a_{n-3}\end{array}\right|}{a_{n-1}},b_2=\frac{-\left|\begin{array}{cc}{a}_n& a_{n-4}\\ a_{n-1}& a_{n-5}\end{array}\right|}{a_{n-1}},\text{etc.}$$ The pattern continues for the $c$-row using the two rows above it, until the $s^0$ row is completed.

Step 2: Apply the Stability Criterion. The system is asymptotically stable (all poles in the LHP) if and only if:

• All coefficients in the first column of the Routh array are positive.
• There are no sign changes in the first column.

The number of sign changes in the first column equals the number of poles located in the right-half s-plane.

Worked Example: Test the stability of a system with characteristic equation $s^4+3s^3+5s^2+4s+2=0$.

Construct the Routh array:

All entries in the first column (1, 3, $\frac{11}{3}$, $\frac{26}{11}$, 2) are positive. There are zero sign changes. Therefore, the system has no RHP poles and is asymptotically stable.

Common Pitfalls

1. Misinterpreting Marginal Stability: Observing a row of all zeros in the Routh array indicates the presence of poles that are symmetric about the origin (e.g., pairs on the imaginary axis). This signals potential marginal stability, but you must complete the auxiliary polynomial from the row above to analyze further. Concluding "stable" at this point is incorrect; you must determine if these symmetric poles are on the imaginary axis (marginally stable) or in the RHP (unstable).
2. Ignoring the "All Coefficients Positive" Pre-check: Before building the Routh array, a necessary (but not sufficient) condition for stability is that all coefficients $(a_n,a_{n-1},...,a_0)$ in the characteristic polynomial are present and positive. If any are missing or negative, you immediately know the system is unstable or marginally stable at best. Skipping this can lead to unnecessary calculations.
3. Calculation Errors in the Array: The recursive calculation of the Routh array is prone to simple arithmetic mistakes. A single error propagates downward, invalidating the entire test. Always double-check your calculations for the first two computed rows ($b$ and $c$ rows) carefully, as errors most commonly originate there.
4. Forgetting About Repeated Imaginary Poles: The basic rule states that simple poles on the imaginary axis yield marginal stability. However, if the Routh test reveals a row of zeros and the auxiliary polynomial has repeated roots on the $j\omega$-axis, the system is actually unstable due to the polynomial growth ($t^n$ terms) in the response. This is a subtle but critical distinction.

Summary

• For an LTI system, asymptotic stability is guaranteed if and only if all poles have negative real parts, placing them strictly in the left-half of the s-plane.
• A single pole with a positive real part (in the right-half plane) makes the entire system unstable, leading to unbounded output growth.
• Purely imaginary, simple poles result in marginally stable behavior, characterized by sustained, undamped oscillations.
• The Routh-Hurwitz criterion is a powerful algebraic tool that determines the number of right-half-plane poles by examining the sign changes in the first column of the Routh array, eliminating the need to calculate the poles explicitly.
• A successful Routh-Hurwitz test requires vigilance for special cases, such as a full row of zeros (indicating symmetric poles) and the essential pre-check that all polynomial coefficients are positive.