Computing Library › Control Theory
Control Theory

Gain and Phase Margins

Gain and phase margins quantify how much loop gain or phase lag a stable feedback system can absorb before it becomes unstable.

Measuring the distance to instability

A feedback loop goes unstable when the open-loop transfer function reaches minus one, magnitude one and phase minus 180 degrees, at some frequency. Gain and phase margins measure how far the loop is from that critical point. They are read directly from a Bode or Nyquist plot and are the classical currency of robustness for single-loop systems.

The two margins

Kronos motion — control room

The gain margin is the factor by which the loop gain can increase before instability, measured at the phase-crossover frequency where the phase is minus 180 degrees: it is the reciprocal of the magnitude there, usually quoted in decibels. The phase margin is the additional phase lag the loop can tolerate before instability, measured at the gain-crossover frequency where the magnitude is one: it is 180 degrees plus the phase there. Positive margins indicate stability with room to spare.

Interpretation and limits

Rules of thumb call for a phase margin of 30 to 60 degrees and a gain margin of two to three, roughly six to ten decibels. Phase margin also relates to transient behavior: larger phase margin generally means less overshoot and more damping. However, the two margins can be misleading for systems with complex loop shapes, where the Nyquist plot approaches minus one obliquely; there the single number can look healthy while a small combined gain-and-phase perturbation destabilizes the loop.

A more complete single measure is the vector margin, or disk margin, the shortest distance from the Nyquist curve to minus one, which captures simultaneous gain and phase perturbations and equals the reciprocal of the peak of the sensitivity function.

For a design-stage single loop, gain and phase margins give a quick, interpretable robustness check, and are the starting point for loop shaping. For multivariable loops, singular-value and mu-based margins generalize them.