In control engineering, system stability stands as the foremost design requirement. Without stability, a control system cannot maintain a desired output, may oscillate uncontrollably, or even damage equipment. Two cornerstone concepts that enable engineers to assess and guarantee stability are the characteristic equation roots and the Routh-Hurwitz criterion. The characteristic equation, derived from a system's transfer function, directly encodes stability information through the location of its roots in the complex plane. The Routh-Hurwitz criterion offers a powerful algebraic method to determine whether all roots lie in the left half-plane—and thus whether the system is stable—without solving for the roots explicitly. Understanding the deep connection between these two tools is essential for every control engineer.

The Characteristic Equation: Foundation of Stability Analysis

Every linear time-invariant (LTI) control system can be represented by a transfer function G(s) = N(s) / D(s), where D(s) is the denominator polynomial. The characteristic equation is defined by setting the denominator polynomial equal to zero:

D(s) = an sn + an-1 sn-1 + ... + a1 s + a0 = 0

This equation originates from the closed-loop system's differential equations or directly from the transfer function denominator. For example, a second-order system common in motion control might yield:

s² + 2ζωn s + ωn² = 0

where ζ is the damping ratio and ωn the natural frequency. The coefficients of the characteristic polynomial contain all the information needed to determine the system’s transient response and stability.

Order and Complexity

The order of the characteristic equation equals the highest power of s and matches the number of energy storage elements in the system (capacitors, inductors, masses, springs, etc.). Real-world systems often have orders ranging from second to tenth or higher. As the order increases, direct root calculation becomes computationally expensive or algebraically prohibitive, making indirect stability tests like Routh-Hurwitz indispensable.

Roots of the Characteristic Equation: The True Stability Indicators

The solutions of the characteristic equation are the poles of the closed-loop system. Their locations in the complex s-plane dictate whether the system is stable, marginally stable, or unstable. The fundamental rule is:

  • All roots with negative real parts – the system is asymptotically stable. Any disturbance decays to zero over time.
  • At least one root with positive real part – the system is unstable. Disturbances cause unbounded growth.
  • Roots on the imaginary axis (zero real part) and no roots in the right half-plane – the system is marginally stable. Sustained oscillations occur.

Roots appear as real values or complex conjugate pairs. A real root corresponds to a first-order exponential mode; a complex conjugate pair corresponds to an oscillatory mode whose frequency is given by the imaginary part and whose decay (or growth) rate is determined by the real part.

Examples

Consider a characteristic equation (s + 1)(s + 2) = 0. Roots are -1 and -2, both with negative real parts → stable.

Consider s² + 4 = 0 (i.e., s = ±2j). Roots lie on the imaginary axis → marginally stable, constant oscillation at frequency 2 rad/s.

Consider (s - 1)(s + 3) = 0. One root is +1 → unstable despite the other root being stable.

Thus, even a single root in the right half-plane renders the entire system unstable. This "all-or-nothing" nature makes a quick stability test incredibly valuable.

The Routh-Hurwitz Criterion: Stability Without Root Calculation

Developed independently by Edward John Routh in 1874 and Adolf Hurwitz in 1895, the Routh-Hurwitz criterion is a systematic algebraic technique that determines whether all roots of a real polynomial have negative real parts. It does so by examining the coefficients of the characteristic polynomial without solving for the roots explicitly.

Constructing the Routh Array

Given a characteristic polynomial in standard form with positive leading coefficient (an > 0):

an sn + an-1 sn-1 + an-2 sn-2 + ... + a0 = 0

Arrange the first two rows of the array as follows:

Row sn: an, an-2, an-4, ...
Row sn-1: an-1, an-3, an-5, ...

Subsequent rows are computed using the determinant formula:

b1 = (an-1 an-2 - an an-3) / an-1
b2 = (an-1 an-4 - an an-5) / an-1

Continue until a single row remains. A row of all zeros or a zero entry in the first column requires special handling (discussed below).

Interpretation of the Routh Array

The number of roots with positive real parts equals the number of sign changes in the first column of the Routh array. If all entries in the first column are positive (no sign changes), then all roots have negative real parts, and the system is stable.

For example, consider the polynomial s³ + 2s² + 3s + 4 = 0. The Routh array:

Row s³: 1, 3
Row s²: 2, 4
Row s¹: (2*3 - 1*4)/2 = 1
Row s⁰: 4

First column: 1, 2, 1, 4 – all positive → no roots in right half-plane → stable. Indeed, the roots are approximately -1.65 and -0.18 ± 1.55j, all with negative real parts.

Special Cases in the Routh Array

A zero in the first column but not an entire row of zeros: Replace the zero with a small positive epsilon (ε) and continue. The sign changes are determined after taking the limit as ε → 0. This situation indicates an imaginary-axis root or a sign change.

A row of all zeros: This occurs when there is an even polynomial factor, typically indicating roots symmetric about the origin (pairs of real roots opposite sign, or imaginary axis roots). Replace the zero row with coefficients from the derivative of the auxiliary polynomial formed from the previous row. The number of roots on the imaginary axis or in the right half-plane can be determined.

The Core Connection: How Routh-Hurwitz Reveals Root Locations

The power of the Routh-Hurwitz criterion lies in its direct algebraic link to the root locations. The criterion does not merely give a yes/no answer; it provides the exact number of roots with positive real parts. This is because the sign changes in the first column correspond to the number of times the polynomial's value changes sign along the imaginary axis as a function of frequency. More formally, the criterion is equivalent to evaluating the Cauchy index of the rational function formed by the polynomial's odd and even parts.

In essence, the Routh-Hurwitz criterion is a polynomial stability test that works by examining the signs of minors of the Hurwitz matrix. For a polynomial to be stable (all roots in the left half-plane), all leading principal minors of the Hurwitz matrix must be positive. The Routh array is a computationally efficient way to check these minors.

The connection can be summarized as:

  • Root locations on the left half-plane → All first column entries positive.
  • Roots on the right half-plane → Sign changes appear, exactly equal to the number of such roots.
  • Imaginary-axis roots → Lead to rows of zeros or zero first column entries.

Thus, the Routh-Hurwitz criterion provides a complete picture of root distribution in the open right half-plane without ever computing a single root. This is invaluable when analyzing high-order polynomials or when designing controllers with variable parameters.

Relation to the Nyquist Criterion and Root Locus

The Routh-Hurwitz criterion is a frequency-domain stability test, but unlike the Nyquist criterion, it does not require plotting. It is purely algebraic. The Nyquist criterion, based on the principle of argument, provides stability margins and can handle time delays. The root locus method, on the other hand, shows how root locations move with gain. The Routh-Hurwitz criterion is often used to validate the stability range of gains obtained from root locus analysis. For example, if a closed-loop characteristic equation has a parameter K, the Routh-Hurwitz criterion can determine the range of K for which the system remains stable.

Practical Applications in Control Engineering

In real-world control system design, engineers rarely compute roots by hand for polynomials above order 3. Instead, they rely on the Routh-Hurwitz criterion to:

  • Determine acceptable ranges for controller gains (e.g., proportional gain Kp in a PID controller). By forming the closed-loop characteristic equation as a function of Kp, the Routh array yields inequalities that define stability boundaries.
  • Check stability of cascaded systems where the overall transfer function denominator is the product of individual denominators.
  • Validate simulation results – a quick algebraic check confirms whether observed oscillations are due to actual instability or numerical artifacts.
  • Teach fundamental stability concepts – because the criterion links coefficient signs to root locations, it reinforces understanding of the relationships between polynomial coefficients and dynamics.

Example: PID Controller Stability

Consider a plant with transfer function G(s) = 1 / (s³ + 2s² + 3s + 4) and a proportional controller C(s) = K. The closed-loop characteristic equation becomes s³ + 2s² + 3s + (4+K) = 0. Using the Routh array:

Row s³: 1, 3
Row s²: 2, 4+K
Row s¹: (6 - (4+K))/2 = (2 - K)/2
Row s⁰: 4+K

For stability, all first column entries must be positive: 1 > 0 always; 2 > 0; (2 - K)/2 > 0 ⇒ K < 2; and 4+K > 0 ⇒ K > -4. Thus, the system is stable for -4 < K < 2. This simple analysis, done without solving a cubic equation, directly gives the gain margin.

Limitations and Extensions

While the Routh-Hurwitz criterion is a stalwart tool, it has limitations:

  • Limited to LTI continuous-time systems – It does not directly apply to discrete-time systems (where the unit circle boundary applies instead of the imaginary axis). However, a bilinear transformation can map the discrete problem.
  • Does not provide exact root locations or transient response – Only stability (presence of right half-plane roots) is assessed. For damping ratios and natural frequencies, root calculation or root locus is required.
  • Numerical issues for high-order polynomials – Coefficient rounding can cause errors. In such cases, polynomial scaling or symbolic computation helps.
  • Cannot handle time delays directly – Time delays introduce infinite-dimensional terms (e.g., e-sT), making the characteristic equation not a polynomial. The Pade approximation can convert the delay into a rational polynomial, but accuracy degrades.

Extensions of the Routh-Hurwitz criterion include:

  • Routh’s algorithm for relative stability – By shifting the imaginary axis leftward (substituting s = z - σ), engineers can determine if all roots have real parts less than -σ, a measure of relative stability.
  • Hurwitz matrix and minors – Direct evaluation of the Hurwitz determinants can be used, but the Routh array is computationally simpler.
  • Discrete-time analogue – The Jury stability test performs a similar role for z-domain polynomials.

Conclusion

The characteristic equation roots and the Routh-Hurwitz criterion are two sides of the same coin in control stability analysis. The roots provide a direct physical interpretation—each pole corresponds to a natural mode of the system—while the Routh-Hurwitz criterion offers an efficient algebraic shortcut to verify that all those poles lie in the stable left half-plane. Together, they form an indispensable duo for any control engineer. By mastering the construction and interpretation of the Routh array and understanding its relationship to the characteristic equation, engineers can quickly assess and design stable control systems, even when explicit root calculation is impractical. Whether you are tuning a PID controller, analyzing a high-order system, or teaching the fundamentals of control, the connection between these two concepts remains a cornerstone of the discipline.

For further reading, refer to the classic textbooks on control systems such as Ogata's Modern Control Engineering or Nise's Control Systems Engineering. Online resources like the Wikipedia article on Routh–Hurwitz stability criterion and Control Global’s control fundamentals provide additional practical insights.