CoursesMEC524 Control SystemsCheat sheetsChapter 4

Chapter 4 cheat sheet

Chapter 4 cheat sheet: System performance analysis

One page for Chapter 4, System Performance Analysis. Print it (PDF)

A model says how a machine moves; a specification says how it should. This chapter reads time constants, overshoot, settling time and steady error off a transfer function, decides with Routh whether a loop is stable, and shows how the poles move as the gain is turned up.

Start here

Find the poles first. A single real pole behaves like a first-order lag; a complex pair gives ζ and ωn, and from them the specifications. Check stability before any specification means anything.

One picture

The step response of a second-order system and the numbers that describe it. The animation swings the poles round a circle of fixed ωn as ζ rises, and the overshoot and settling time follow.

The step response of a second-order system and the numbers that describe it. The animation swings the poles round a circle of fixed ω_n as ζ rises, and the overshoot and settling time follow.
ω_n = 2 fixed, ζ 0.1 → 0.9: poles slide round the circle; %OS, T_p, T_s update

Which formula?

Read the situation first, then pick the formula. The numbers in brackets are the handbook's equation numbers.

  1. One time constant

    G(s) = (Y(s))/(U(s)) = K/(τ s + 1)(4.2)

    τ is the time to 63.2 % of the final value.

    y(t) = K(1 − e^(−t/τ)), t ≥ 0(4.3)

    The unit-step response.

  2. A second-order plant

    G(s) = (Y(s))/(U(s)) = (K ω_n ^2)/(s^2 + 2ζω_n s + ω_n ^2)(4.6)

    Standard form: read ζ and ωn off the denominator.

    s_1,2 = −ζω_n ± ω_n √(ζ^2 − 1)(4.7)

    Complex for ζ<1, real and repeated at ζ=1.

  3. How much it overshoots

    M_p = e^(−πζ/√(1−ζ^2)) × 100%§4.2

    Depends on ζ alone.

  4. How long until it settles

    t_s = 4/ζω_n§4.2

    The 2 % band; set by the real part of the poles.

  5. The error left at the end

    e_ss = lim_t→∞ e(t) = lim_s→ 0 s E(s) = lim_s→ 0 (s R(s))/(1 + G(s))(4.10)

    Unity feedback, stable loop; the system type decides which inputs it tracks.

  6. Is the loop stable?

    b_1 = (a_n−1 a_n−2 − a_n a_n−3 )/(a_n−1 ), b_2 = (a_n−1 a_n−4 − a_n a_n−5 )/(a_n−1 ), …(4.12)

    Routh array: each sign change in the first column is one right-half-plane pole.

  7. Where the poles go as K grows

    θ_a = ((2k+1) × 180°)/(n − m), k = 0, 1, …, n − m − 1(4.17)

    Asymptote angles, for n−m branches going to infinity.

    σ_a = (Σ_j p_j − Σ_i z_i )/(n − m)(4.18)

    Where the asymptotes meet the real axis.

Reading the pictures

The poles sit on a circle of radius ω_n ; the angle from the negative real axis sets ζ.
The poles sit on a circle of radius ωn; the angle from the negative real axis sets ζ.
The stability map: left half-plane decays, right half-plane grows, the imaginary axis rings forever.
The stability map: left half-plane decays, right half-plane grows, the imaginary axis rings forever.
A root locus: the branches leave the open-loop poles and two of them cross into the right half-plane at the critical gain.
A root locus: the branches leave the open-loop poles and two of them cross into the right half-plane at the critical gain.

Ways to lose marks

  • Specifications need stability. A step-response specification means nothing for an unstable loop: check every pole first.
  • Two different knobs. Overshoot depends on ζ only; settling time on ζωn.
  • A zero in the Routh column. Replace it by a small ε, or use the auxiliary polynomial for a whole row of zeros.
  • System type. Count the integrators in the open loop G(s), not in the closed loop.
  • Asymptotes. They need n−m: open-loop poles minus open-loop zeros.

Handbook: § 4.0–§ 4.6 for the worked examples and tutorials. Slides: the Chapter 4 deck, with every animation. Download the one-page PDF to print.