Compensation Techniques

Compensation techniques in control systems improve system performance by modifying system dynamics.

Drafted with Aria, reviewed by the AiCanCode.org team. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Compensation modifies loop gain and phase to meet tracking, disturbance rejection and stability-margin requirements. A compensator must be evaluated together with its plant and feedback path.

Lead and lag

For G_c(s) = K_c(s+z)/(s+p), with positive z and p, the zero lies at −z and pole at −p. Lead requires p > z; lag requires z > p. Phase lead can increase phase margin near crossover. Lag can increase low-frequency loop gain relative to crossover to improve tracking, but its added phase lag must be included in stability checks.

A normalized lead network is C(s) = K(1+Ts)/(1+αTs), with 0 < α < 1. Its maximum phase is φ_max = asin((1−α)/(1+α)), at ω_m = 1/(T√α). Its magnitude at that frequency is K/√α. Changing the magnitude shifts crossover, so adding φ_max does not automatically increase the final phase margin by exactly that amount.

Worked example

Design a lead network providing maximum phase lead 30° at 10 rad/s and unity magnitude at that frequency.

α = (1−sin30°)/(1+sin30°) = 1/3. T = 1/(10√(1/3)) = 0.173205 s. Choose K = √α = 0.577350 so |C(j10)| = 1.

C(s) = 0.577350(1+0.173205s)/(1+0.057735s). Its zero is at −5.7735 rad/s and pole at −17.3205 rad/s. At 10 rad/s the phase is atan(1.73205)−atan(0.57735) = 60°−30° = 30°, and magnitude is one.

This completes the network design. It is not a claim about an unspecified plant’s phase margin. For an existing loop with unity magnitude at 10 rad/s, this network retains that crossing; check for any additional crossings and verify overall closed-loop stability.

Design procedure

Determine the uncompensated loop’s actual crossover and margins. Choose a target crossover consistent with bandwidth and noise limits. Place lead or lag poles and zeros, adjust gain using the magnitude condition, then recompute all crossings and closed-loop poles. Check time response, actuator limits and model uncertainty.

Common mistakes

Calling (s+1)/(s+0.1) a lead network; interpreting z and p as the signed pole locations; assuming a phase margin without finding a unity-gain crossing; and neglecting crossover movement after compensation.

Quick check

  1. Is (s+1)/(s+10) lead or lag?
  2. Is (s+10)/(s+1) lead or lag?
  3. Does 30° maximum network lead guarantee 30° improvement in every loop’s phase margin?

Answers: 1. Lead. 2. Lag. 3. No; the crossover and complete loop must be checked.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?