Lag Compensator Design

Core Idea

Lag compensator design improves steady-state accuracy by increasing the low-frequency loop gain without significantly altering the gain crossover frequency or phase margin. The compensator C(s) = K_c · (s + z_c)/(s + p_c) with z_c > p_c (zero farther from origin than pole) provides a gain increase of z_c/p_c = β at frequencies well below z_c while contributing negligible magnitude change near the crossover frequency. The design procedure is: (1) set the gain K_c to meet the transient response specification (desired crossover frequency and phase margin) as if no lag network were present; (2) compute the improvement factor β needed to meet the steady-state error specification (e.g., β = K_v,required/K_v,current for a velocity error constant); (3) place the zero z_c well below the gain crossover frequency (typically one decade or more below ωgc) to avoid contributing negative phase at crossover; (4) set p_c = z_c/β. The lag compensator's negative phase contribution near its corner frequencies is kept harmless by placing both z_c and p_c at low frequencies far from ωgc. The result is improved steady-state performance with minimal impact on the transient response already established by the gain selection.

How It's Best Learned

Design a lag compensator for a unity-feedback system where the uncompensated gain meets phase margin requirements but the velocity error constant K_v is too low by a factor of 10. Walk through the β calculation, zero/pole placement, and verify on the Bode plot that the phase margin is preserved while the low-frequency gain increases by 20 dB. Compare step and ramp responses before and after compensation to see the steady-state error reduction directly.

Common Misconceptions

• A lag compensator does not add a pole at the origin and therefore does not change the system type — it increases the error constant (Kp, Kv, or Ka) by a finite factor β, but a Type 0 system remains Type 0 with a finite steady-state error to a step input.
• The lag compensator does contribute negative phase (up to −90° between p_c and z_c), which can erode phase margin if the corner frequencies are placed too close to the gain crossover frequency — the "place it a decade below ωgc" rule is a practical necessity, not a minor detail.
• Lag compensation can slow down the transient response because the closed-loop system acquires a slow pole-zero pair near the origin, producing a long-duration, low-amplitude tail in the step response that can extend the effective settling time.

Explainer

The fundamental challenge the lag compensator addresses is this: you have already chosen a gain K that places the gain crossover frequency at the desired location, giving you the phase margin (and therefore the transient response) you want. But when you compute the steady-state error, you find that the error constant (Kv for ramp inputs, Kp for step inputs) is too small by a factor β — the system tracks accurately enough in speed, but not precisely enough in steady-state. You need more low-frequency gain without disturbing the crossover region. That is exactly what a lag compensator provides.

The compensator C(s) = K_c · (s + z_c)/(s + p_c) with z_c > p_c has a simple frequency-domain interpretation. At frequencies well below both corner frequencies (ω ≪ p_c < z_c), the numerator and denominator both contribute approximately their DC values, and the DC gain of the compensator is z_c/p_c = β > 1. At frequencies well above both corner frequencies (ω ≫ z_c > p_c), numerator and denominator magnitudes cancel and the gain returns to 1. So the lag compensator is essentially a low-frequency gain booster: it multiplies the loop gain by β at DC and low frequencies, where steady-state accuracy is determined, while leaving the high-frequency Bode plot (including the crossover region) nearly unchanged.

The catch is that between the two corner frequencies, the phase response dips negative — up to -90° at the geometric mean of p_c and z_c. This is why placement matters so much: if z_c is close to the gain crossover frequency ωgc, that negative phase dip erodes the phase margin you carefully designed. The rule of thumb — place z_c at least one decade below ωgc, i.e., z_c ≤ ωgc/10 — ensures the phase contribution at crossover is small (roughly -6° or less). Then set p_c = z_c/β to achieve the target gain improvement.

Concretely: suppose you need Kv = 20 but the uncompensated system delivers Kv = 2 at the desired crossover frequency. You need β = 10 (a 20 dB low-frequency gain boost). If ωgc = 10 rad/s, place z_c = 1 rad/s and p_c = 0.1 rad/s. The compensator C(s) = (s + 1)/(s + 0.1) multiplies the low-frequency open-loop gain by 10, raises Kv from 2 to 20, and contributes only a small phase dip near ω = 0.3 rad/s — far from the crossover at 10 rad/s. Check the result on the Bode plot: the magnitude is shifted up 20 dB at low frequencies, the phase margin at ωgc is nearly unchanged, and the ramp tracking error has decreased by a factor of 10. The tradeoff is a slow pole-zero pair near the origin that can produce a long, low-amplitude transient tail — acceptable for most applications, but worth checking in the step response.