Introduction to the Routh-Hurwitz Criterion

The Routh-Hurwitz criterion is a cornerstone of classical control theory, providing a rapid algebraic method to assess the stability of a linear time-invariant (LTI) system without explicitly computing the roots of its characteristic polynomial. Developed independently by Edward John Routh in 1876 and Adolf Hurwitz in 1895, the method remains indispensable for engineers who need to guarantee stable, predictable behavior in feedback control systems, electrical networks, and mechanical structures. While modern software can compute poles directly, the Routh-Hurwitz table offers deep insight into the relationship between polynomial coefficients and root locations, making it a critical skill for practical engineering analysis.

This article provides a detailed, step-by-step guide to building and interpreting the Routh-Hurwitz table, with emphasis on real-world application. You will learn how to detect instability from sign changes, handle special cases such as zeros in the first column or entire rows of zeros, and apply the criterion to a variety of engineering domains. By the end, you will be able to quickly evaluate system stability and understand the limitations of the method.

Constructing the Routh-Hurwitz Table

The Routh-Hurwitz table is built from the coefficients of the characteristic equation P(s) = ansn + an-1sn-1 + ... + a1s + a0 = 0, where all coefficients are real numbers and an > 0 by convention (if an is negative, multiply the entire polynomial by -1). The table consists of rows indexed by powers of s, starting from sn down to s0.

Step 1: Arrange the First Two Rows

List the coefficients of the polynomial in alternating rows:

  • Row sn: contains coefficients an, an-2, an-4, … (coefficients of even-indexed terms, starting with the highest degree).
  • Row sn-1: contains coefficients an-1, an-3, an-5, … (coefficients of odd-indexed terms).

Continue until no coefficients remain; fill empty entries with zeros.

Step 2: Compute Subsequent Rows

Each subsequent row (for sn-2, sn-3, …) is calculated using a determinant formula. For an element in row i and column j, the formula is:

bj = -(det[[a1,1, a1,j+1], [a2,1, a2,j+1]]) / a2,1

where a1,1 and a2,1 are the first elements of the two preceding rows. In practice, the pattern is:

  • Take the first element of the previous row (which becomes the denominator).
  • Create a 2×2 matrix using the first column of the two preceding rows and the next columns of those rows.
  • Compute the determinant, change its sign, and divide by the denominator.

For example, to compute the first element of row sn-2 from rows sn and sn-1:

b1 = (an-1 * an-2 - an * an-3) / an-1

Repeat for each column until the row is complete (the number of elements in each row decreases as you go down). Continue until you reach the s0 row, which contains only one element.

Interpreting Stability from the First Column

Once the table is complete, focus exclusively on the first column (the leftmost column of each row). The Routh-Hurwitz stability criterion states:

  • All entries in the first column must be positive (or all negative) for the system to be stable. If any entry is zero or a sign change occurs, the system has roots in the right-half s-plane and is unstable.
  • The number of sign changes in the first column (from positive to negative or vice versa) equals the number of roots with positive real parts. A root on the imaginary axis (marginal stability) appears as a zero in the first column or an entire row of zeros.

Example: Stable System

Consider the polynomial P(s) = s4 + 2s3 + 3s2 + 4s + 5. The Routh-Hurwitz table is constructed as follows:

s rowCoefficients
s41, 3, 5
s32, 4, 0
s2(2*3 - 1*4)/2 = 1, (2*5 - 1*0)/2 = 5, 0
s1(1*4 - 2*5)/1 = -6, (1*0 - 2*0)/1 = 0
s0(-6*5 - 1*0)/(-6) = 5

The first column entries are: 1, 2, 1, -6, 5. There is one sign change (from 1 to -6, then back to 5 counts as two sign changes? Actually only one sign change: positive to negative at the s1 row. Then negative to positive at s0? Wait, careful: The first column values: row s^4: +1, s^3: +2, s^2: +1, s^1: -6, s^0: +5. Sign goes +,+,+,-,+ => there are two sign changes: + to - (between s^2 and s^1), and - to + (between s^1 and s^0). So the system has two roots with positive real parts. This polynomial is indeed unstable.

For a truly stable system, e.g., P(s) = s2 + 2s + 1, the first column would be [1, 2, (2*1 - 1*0)/2 = 1] => all positive => stable.

Common Mistake: Counting Sign Changes

Always start at the top row and move downward. Each transition from positive to negative or negative to positive counts as one sign change. Do not skip rows; include the s0 row.

Practical Engineering Applications

The Routh-Hurwitz criterion is not just a theoretical exercise — it is widely used in design and analysis across multiple engineering disciplines.

Control System Design

In feedback control, the characteristic equation arises from the closed-loop transfer function. Engineers adjust controller gains (e.g., PID parameters) and use the Routh-Hurwitz table to find ranges of gains that keep all poles in the left-half plane. This is especially valuable during the early design phase before simulation, as it provides immediate stability boundaries without solving high-order polynomials. For example, in tuning a proportional controller for a third-order plant, the Routh array reveals the maximum allowable gain before instability occurs.

Aerospace Engineering

Aircraft and spacecraft flight control systems must be stable under all flight conditions. The Routh-Hurwitz criterion is used to assess stability margins for autopilots, stability augmentation systems, and structural mode suppression. Given the complexity of aircraft dynamics (often represented by state-space models of order 10 or higher), the criterion offers a quick check after linearization. It also helps in analyzing the effect of parameter variations (e.g., fuel consumption) on stability.

Electrical Circuits and Power Systems

Stability of electrical networks — from simple op-amp circuits to large-scale power grids — can be evaluated using the characteristic polynomial of the system's transfer function. For instance, the root locus of a power system stabilizer can be informed by Routh-Hurwitz analysis to prevent low-frequency oscillations. Similarly, in switched-mode power supplies, the criterion helps decide compensation network parameters to avoid instability under load changes.

Mechanical Systems and Robotics

Robotic manipulators, vehicle suspensions, and flexible structures often have characteristic equations derived from Newton-Euler or Lagrange equations. The Routh-Hurwitz criterion aids in determining whether a given set of spring/damper constants leads to stable motion. For example, in designing a two-wheeled balancing robot, engineers use the criterion to find the minimum damping coefficient required to stabilize the inverted pendulum mode.

Chemical Process Control

In process industries, controllers for temperature, pressure, and flow must be tuned to avoid oscillations or runaway conditions. The Routh-Hurwitz table is applied to the characteristic equation of a chemical reactor's linearized model, helping to set PID gains that ensure stability despite time delays (which are approximated via Padé polynomials).

Handling Special Cases in the Routh-Hurwitz Table

Two common anomalies occur when constructing the table: a zero in the first column (but not an entire row), and an entire row of zeros. Both require special treatment to continue the analysis.

Zero in the First Column (but Row Not All Zeros)

If the first element of a row is zero while the rest of the row is non-zero, the table cannot be continued with the normal formula because division by zero is required. The standard solution is to replace the zero with a small positive number ε (epsilon) and proceed with the calculations. After completing the table, analyze the signs of the first column as ε → 0+. The number of sign changes as ε approaches zero from the positive side indicates the number of right-half-plane poles. This method is valid because a zero first column often corresponds to a pair of imaginary-axis roots or a marginally stable system.

Example: Polynomial P(s) = s3 + 2s2 + s + 2. The first two rows: s3: [1, 1], s2: [2, 2]. Compute s1: (2*1 - 1*2)/2 = 0/2 = 0. So the first element of s1 row is zero. Replace with ε: s1 = [ε, ?]. Next element in s1 uses columns: (2*2 - 1*0)/2? Wait, careful: The second element of s1 uses the second columns of the two preceding rows. Actually since s2 row has two entries [2,2] and s3 has two entries [1,1], we compute: first element of s1 = (2*1 - 1*2)/2 = 0 → set to ε. Second element of s1 = (2*1 - 1*0)/2 = 2/2 = 1. So s1 row = [ε, 1]. Now compute s0 = (ε*2 - 2*1)/ε = (2ε - 2)/ε = 2 - 2/ε. As ε → 0+, 2/ε → +∞, so 2 - 2/ε → -∞. Thus the first column becomes: s3: 1, s2: 2, s1: ε (positive small), s0: -∞ (negative). Sign changes: from s2 (positive) to s1 (positive) = no change; from s1 (positive) to s0 (negative) = one change. So one right-half-plane pole. Indeed, the polynomial factors as (s+2)(s2+1) giving poles at -2, ±j, so one pair on imaginary axis and one stable pole — the system is marginally stable, not strictly stable. The Routh-Hurwitz table with ε correctly indicates a marginal case.

Entire Row of Zeros

When an entire row of the Routh-Hurwitz table is zero, it signals that the polynomial has symmetrically located roots — roots that are either purely imaginary, real and opposite in sign, or complex-conjugate pairs on the imaginary axis. To continue, construct an auxiliary polynomial from the row immediately above the zero row. The auxiliary polynomial is formed by taking the coefficients from that row and treating the powers of s as decreasing by 2 each term (e.g., for row sk, the polynomial is c1sk + c2sk-2 + c3sk-4 + ...). Differentiate this polynomial with respect to s, and use the coefficients of the derivative to replace the zero row. Then continue the table normally. The roots of the auxiliary polynomial (which are also roots of the original characteristic equation) provide information about the marginal or symmetric behavior.

Example: Polynomial P(s) = s4 + 5s3 + 10s2 + 20s + 24. The table construction yields a row of zeros at s2. The row above (s3) has coefficients [5,20]; the auxiliary polynomial is 5s2 + 20 (note: powers s3 and s1? Actually row s3 corresponds to powers 3,1; the auxiliary polynomial uses the coefficients directly: for row s3 with coefficients [5,20], the polynomial is 5s3 + 20s? Wait, standard method: the coefficients of the auxiliary polynomial come from the row above the zero row, and the powers of s start at the row index and decrease by 2. So if the zero row is s2, the row above is s3. The auxiliary polynomial is 5s3 + 20s? That would be 5s(s2+4) which has roots at s=0, ±j2. But the actual polynomial has roots at -2, -3, ±j2? Let's check: P(s)=(s+2)(s+3)(s2+4)= s^4 +5s^3+10s^2+20s+24. Yes. The auxiliary polynomial from row s^3 (assuming zero row at s^2) is indeed derived from the coefficients of the row s^3: coefficients 5 and 20, with s powers s^3 and s^1. That polynomial is 5s^3+20s. Differentiate: 15s^2+20. Use coefficients [15,20] to replace the zero row. Then continue. The first column will eventually show no sign changes? Actually this system has stable poles at -2,-3 and marginal poles at ±j2, so first column should have no sign changes (because no right-half-plane poles) but a zero in the first column or a sign change? With the derivative row, the table can be completed and the sign pattern confirms stability plus the auxiliary polynomial reveals the marginal roots. The key is to replace the zero row with coefficients of the derivative of the auxiliary polynomial, then proceed normally. This method allows detection of roots on the imaginary axis.

Limitations of the Routh-Hurwitz Criterion

While powerful, the Routh-Hurwitz criterion has several limitations that every engineer should understand:

  • Only applicable to linear time-invariant (LTI) systems with polynomial characteristic equations. It cannot directly handle time delays, nonlinearities, or time-varying parameters.
  • No direct information about transient response (such as settling time, overshoot, damping ratio). The criterion only indicates stability or instability, not performance.
  • Sensitive to coefficient errors — small measurement or rounding errors can lead to incorrect sign evaluations, especially near stability boundaries.
  • Does not quantify stability margins (gain margin, phase margin). For that, frequency-domain methods like Nyquist or Bode plots are needed.
  • Cannot easily handle polynomials with symbolic coefficients beyond simple parameterizations (e.g., a single gain). Systems with many variable parameters become cumbersome.

Despite these limitations, the Routh-Hurwitz table remains a fast, intuitive, and mathematically rigorous first-line tool for stability analysis.

Practical Tips for Using the Routh-Hurwitz Table

  1. Always check the necessary condition first: For stability, all coefficients of the characteristic polynomial must be positive. Any zero or negative coefficient (especially for high-degree terms) immediately signals possible instability — but this is not sufficient; the table is still needed to confirm.
  2. Use numerical precision: When constructing large tables by hand, use symbolic fractions or high-precision arithmetic to avoid sign errors.
  3. Combine with root locus: For design, use the Routh-Hurwitz criterion to find the range of a gain parameter, then use root locus to examine the closed-loop pole migration for transient behavior.
  4. In practice, verify with simulation: After applying the criterion, always simulate the system (e.g., step response) in software like MATLAB, Simulink, or Python Control Library to confirm.
  5. For high-order systems (order > 10), consider using numerical methods (eigenvalue computation) for accuracy, but use the Routh-Hurwitz table for insight into parameter sensitivity.

External Resources and Further Reading

For a deeper understanding of the Routh-Hurwitz criterion and its place in control theory, the following authoritative resources are recommended:

Conclusion

Mastering the Routh-Hurwitz table empowers engineers to quickly assess system stability during the early design phase, without resorting to complex root-solving algorithms. By understanding how to construct the table, interpret sign changes, and handle special cases (epsilon method and auxiliary polynomial), you gain a reliable, intuitive tool that works for polynomials of any order. The criterion's direct connection to coefficient patterns makes it especially valuable for parametric studies, gain tuning, and stability margin estimation in a wide array of engineering disciplines — from aerospace and automotive control to power electronics and process automation. While modern computational tools have reduced the need for manual table construction, the conceptual insight provided by the Routh-Hurwitz criterion remains an essential part of every control engineer's analytical toolkit.