Computing Library › Control Theory
Control Theory

The Small-Gain Theorem

The small-gain theorem guarantees closed-loop stability when the product of the gains of two interconnected systems is less than one.

Statement

Consider two stable systems connected in a feedback loop. If each has finite gain, meaning it maps bounded-energy inputs to bounded-energy outputs with a finite ratio, and the product of their gains is strictly less than one, then the interconnection is stable and its gain is bounded. The result holds in very general settings, including nonlinear and time-varying systems, because it depends only on input-output gains, not on internal structure.

Robustness interpretation

Kronos motion — closed loop

The theorem is the engine of robust stability. Pull the uncertainty out of a control loop into a perturbation block Delta connected to a nominal system M. If M is stable with H-infinity norm below one over gamma, then the loop is stable for every stable Delta with norm below gamma. This is why minimizing the H-infinity norm buys guaranteed robustness against norm-bounded uncertainty.

Conservatism

Because it uses only a scalar gain and ignores phase and structure, the small-gain condition is sufficient but not necessary, and it can be conservative. When the uncertainty is structured, the structured singular value gives a tighter, exact test. When phase information matters, passivity-based tools do better.

Refinements sharpen it. The integral quadratic constraint framework generalizes small gain by allowing frequency-dependent multipliers that capture more about the perturbation than a single number, recovering much of the lost tightness while keeping a convex test.

For a design-stage magnet loop with a poorly known high-frequency mode, the small-gain theorem gives a quick, rigorous sufficient condition: keep the loop gain below the inverse of the modeled uncertainty size across the uncertain band. It is the first robustness check most engineers reach for.