System stability tells engineers whether a circuit, aircraft controller, robot, or chemical process will return to a desired condition after a disturbance. For a linear time-invariant system, stability is determined by the locations of the poles of its transfer function in the complex s-plane. Poles in the left half-plane produce decaying responses, while poles in the right half-plane produce growing responses.
This makes pole location one of the most important links between algebra, graphs, and physical behavior.
Understanding Engineering: System Stability and Routh-Hurwitz
The Routh array turns a characteristic polynomial into a structured calculation. It is useful because finding every root becomes difficult for fourth-order or higher models, especially when a controller gain is still being chosen. Start by writing the closed-loop characteristic polynomial in descending powers.
The first two rows use alternating coefficients from that polynomial. Each lower row is calculated from the two rows above it. The arithmetic uses products, subtraction, and division, so a small copying error can change the result.
Before starting, engineers usually make the leading coefficient positive. Multiplying the whole polynomial by a positive constant does not affect the stability result.
The first column gives a fast map of the unstable part of the system. Each reversal of sign represents one mode that grows rather than dies away. This is more informative than a simple stable or unstable label.
For example, two sign reversals mean there are two unstable roots. A design can then be adjusted while watching the array change. When a first-column entry approaches zero, the system is approaching a stability boundary.
In feedback control, increasing gain often improves tracking at first, but excessive gain can bring the system to this boundary. The Routh test identifies the limit, though it does not say how comfortably inside the stable region the design sits.
Two special cases need careful handling. If the first entry of a row is zero while other entries in that row remain, engineers temporarily replace the zero with a very small positive number. They then inspect the signs as that number approaches zero.
If an entire row becomes zero, the previous row forms an auxiliary polynomial. Differentiating that polynomial provides the replacement row. This case often signals roots arranged symmetrically about the origin, including possible oscillatory roots.
Such a result deserves extra attention. An ideal model may predict constant-amplitude oscillation, but friction, saturation, sensor noise, delay, and component variation can make the real system behave poorly.
Students often meet this method when studying a motor speed controller, an active circuit, or a process temperature loop. The polynomial comes from combining the plant model with the controller and setting the closed-loop denominator equal to zero. Keep the physical meaning connected to the algebra.
Large controller gains can demand unrealistic motor voltage. A stable chemical process can still respond too slowly for safe operation. Routh-Hurwitz applies directly to continuous-time models.
Sampled digital controllers use a different stability region, based on the unit circle rather than this test. When learning the method, practice setting up rows neatly, checking coefficient order, and treating zero-row cases as important physical warnings rather than awkward calculation exceptions.
Key Facts
- A continuous-time LTI system is asymptotically stable if all poles have negative real parts.
- For characteristic equation a_n s^n + a_(n-1) s^(n-1) + ... + a_0 = 0, all coefficients must have the same sign for stability, but this is not sufficient.
- Routh-Hurwitz criterion: the number of sign changes in the first column of the Routh array equals the number of right half-plane poles.
- For a second-order polynomial a s^2 + b s + c, stability requires a > 0, b > 0, and c > 0.
- For s^3 + a s^2 + b s + c, stability requires a > 0, b > 0, c > 0, and ab > c.
- Pure imaginary poles indicate marginal stability only when they are simple and no poles lie in the right half-plane.
Vocabulary
- Pole
- A pole is a value of s that makes the transfer function denominator equal to zero and strongly shapes the system response.
- s-plane
- The s-plane is the complex plane used in control engineering, with horizontal real axis sigma and vertical imaginary axis j omega.
- Asymptotic stability
- Asymptotic stability means the natural response decays to zero as time goes to infinity.
- Routh array
- A Routh array is a tabular arrangement of characteristic equation coefficients used to count right half-plane poles without solving for the roots.
- Marginal stability
- Marginal stability means the response remains bounded but does not decay to zero, often due to simple poles on the imaginary axis.
Common Mistakes to Avoid
- Assuming positive coefficients guarantee stability is wrong because higher-order polynomials can have all positive coefficients and still have right half-plane poles.
- Counting any negative entry in the Routh table instead of sign changes is wrong because the stability test uses changes of sign only in the first column.
- Ignoring a zero in the first column is wrong because it can make the next row undefined and requires the epsilon method or a special case procedure.
- Calling every imaginary-axis pole stable is wrong because repeated imaginary-axis poles or any right half-plane pole make the system unstable.
Practice Questions
- 1 Use the Routh-Hurwitz criterion to determine the number of right half-plane poles for s^3 + 2s^2 + 3s + 10 = 0.
- 2 Construct the Routh array for s^4 + 3s^3 + 5s^2 + 4s + 2 = 0 and decide whether the system is stable.
- 3 A system has poles at -2, -1 + 3j, -1 - 3j, and 0 + 4j. Explain whether the system is asymptotically stable, marginally stable, or unstable.