Skip to content
Automatic Controls 1

Solved root locus exercises, with the loci drawn

Nine exercises with solutions: gain for a damping, asymptotes and breakaway, circular locus, departure angles, stabilising an unstable process, stabilising zero, locus with respect to a parameter, design for overshoot.

Root locus exercises of the kind set in Control Systems courses, in three levels: basic, with second-order systems and gain computation; intermediate, with asymptotes, breakaway points, departure angles and imaginary-axis crossings; exam, with unstable processes, compensation with a zero, locus with respect to a parameter and design to a specification. The result is given straight away; the full solution, with the locus drawn, opens with a click. The proved rules and the MATLAB and Python code are in the in-depth study linked at the top.

Basic level

Exercise 1. The simplest locus

Sketch the locus of L(s) = \dfrac{K}{s(s + 2)} and find the gain for which the closed-loop poles have damping \zeta = 0.707.

Result: branches on the real axis between 0 and -2, breakaway at -1 for K = 1, then the vertical line \text{Re}\,s = -1; K = 2.

Solution

Characteristic equation: s^2 + 2s + K = 0. Real axis: the stretch between 0 and -2 (to the right of each of its points there is one pole). Two asymptotes at \pm 90° from the centroid \frac{0 - 2}{2} = -1.

Breakaway: \frac{d}{ds}(s^2 + 2s) = 2s + 2 = 0, s = -1, with K = 1. For K \gt  1 the poles are -1 \pm i\sqrt{K - 1}: the real part stays at -1 (the sum of the poles is constant, rule 7).

Damping: comparing with s^2 + 2\zeta\omega_ns + \omega_n^2, \omega_n = \sqrt K and \zeta = \frac{1}{\sqrt K}. For \zeta = 0.707: K = 2, poles -1 \pm i. Increasing the gain makes the response more oscillatory but not faster to settle: the settling time 4/\zeta\omega_n = 4 s does not change.

Root locus of K over s times s plus 2
The branches start from 0 and minus 2, meet at minus 1 and go up vertically; the dashed green line at 45 degrees corresponds to damping 0.707 and meets the locus at minus 1 plus or minus i.

Exercise 2. A gain for an overshoot

For L(s) = \dfrac{K}{s(s + 2)(s + 4)} find the gain that gives the dominant poles a damping of 0.5, the position of the third pole and the critical gain.

Result: K = 8.3; dominant poles -0.667 \pm 1.155i, third pole -4.67; K_{crit} = 48.

Solution

The dominant poles lie on the line s = r\,e^{\pm i120°} (damping 0.5, angle 60° from the negative half-axis). By rule 7, the sum of the three poles is always -6: if the dominant ones are -a \pm ia\sqrt 3, the third is -6 + 2a.

Requiring (s^2 + 2as + 4a^2)(s + 6 - 2a) = s^3 + 6s^2 + 8s + K, the coefficient of s gives 4a^2 + 2a(6 - 2a) = 8, i.e. 12a = 8, a = 0.667. Then the dominant poles are -0.667 \pm 1.155i, the third -4.67, and K = 4a^2(6 - 2a) = 1.778\cdot 4.667 = 8.30.

Critical gain (Routh, or s = i\omega): 6\cdot 8 = K, K = 48, \omega = 2\sqrt 2. The gain margin with K = 8.3 is 48/8.3 = 5.8.

The third pole is seven times further from the axis than the dominant ones: its influence is small, and the actual overshoot (15.5 %) is close to that of the second-order system (16.3 %).

The locus crossed by the damping line 0.5
The green dots are the three closed-loop poles for K equal to 8.3: the two dominant ones on the dashed line and the third on the real axis at minus 4.67.

Intermediate level

Exercise 3. Four poles: asymptotes, breakaway, stability

For L(s) = \dfrac{K}{s(s + 1)(s + 3)(s + 4)} find the asymptotes, the breakaway points and the critical gain.

Result: centroid -2, angles \pm 45° and \pm 135°; breakaway at -0.419 and -3.581 (both with K = 2.25); K_{crit} = 26.25 with \omega = 1.22.

Solution

Real axis: stretches [-1, 0] and [-4, -3]. Asymptotes: n - m = 4, centroid \frac{0 - 1 - 3 - 4}{4} = -2, angles 45°, 135°, 225°, 315°.

Breakaway: D = s^4 + 8s^3 + 19s^2 + 12s, D' = 4s^3 + 24s^2 + 38s + 12 = 0: roots -0.419, -2, -3.581. The point -2 is not on the locus (between -1 and -3 there are two poles to the right, an even number). At the other two K = |D| = 2.25: the same K by symmetry (the poles are symmetric about -2).

Routh on s^4 + 8s^3 + 19s^2 + 12s + K: second row 8, 12; third \frac{8\cdot 19 - 12}{8} = 17.5 and K; fourth \frac{17.5\cdot 12 - 8K}{17.5}, positive for K \lt  26.25. Auxiliary equation 17.5s^2 + 26.25 = 0: \omega = 1.22.

Locus with four poles and its asymptotes
The pairs of branches leave the axis at minus 0.42 and minus 3.58 and approach the asymptotes at 45 and 135 degrees starting from minus 2; the right-hand branches cross the imaginary axis.

Exercise 4. The circular locus and maximum damping

For L(s) = K\dfrac{s + 3}{s(s + 1)} find the breakaway points and the gain that gives maximum damping of the complex poles.

Result: breakaway at -0.551 and -5.449; maximum damping \zeta = 0.577 for K = 1, with poles -1 \pm i\sqrt 2.

Solution

Breakaway (in-depth study): s^2 + 6s + 3 = 0, s = -3 \pm\sqrt 6.

The complex part of the locus is the circle centred at -3 with radius \sqrt 6. The damping of a pole is the cosine of the angle between the line to the origin and the negative half-axis: it is maximum where the line from the origin is tangent to the circle. There \sin\theta = \frac{\sqrt 6}{3}, \zeta = \cos\theta = \sqrt{1 - \frac{6}{9}} = 0.577, and the distance from the origin is \sqrt{9 - 6} = \sqrt 3.

The pole is \sqrt 3\,e^{i(180° - 54.7°)} = -1 + i\sqrt 2. From the equation s^2 + (1 + K)s + 3K = 0, the real part is -\frac{1 + K}{2} = -1, so K = 1. Check: s^2 + 2s + 3 has roots -1 \pm i\sqrt 2.

Circular locus around the zero
The circle centred at minus 3 with radius root 6 joins the two breakaway points; the point of maximum damping is where the line from the origin touches it.

Exercise 5. Departure angle and crossing

For L(s) = \dfrac{K}{(s + 3)(s^2 + 2s + 2)} compute the departure angle from the pole -1 + i, the asymptotes and the critical gain.

Result: 63.4°; centroid -1.67, angles \pm 60°, 180°; K_{crit} = 34 with \omega = 2.83.

Solution

Departure (rule 6): \theta = 180° - \arg(p + 3) - \arg(p - \bar p) = 180° - 26.6° - 90° = 63.4°.

Asymptotes: \sigma_a = \frac{-3 - 1 - 1}{3} = -1.67.

Characteristic equation s^3 + 5s^2 + 8s + 6 + K = 0. Routh: 5\cdot 8 \gt  6 + K, i.e. K \lt  34. For K = 34: 5s^2 + 40 = 0, \omega = \sqrt 8 = 2.83.

Check with rule 7: for K = 34 the poles are \pm 2.83i and the third must be -5 (sum -5): indeed (s + 5)(s^2 + 8) = s^3 + 5s^2 + 8s + 40. ✓

Departure from the complex poles
The branch leaves the pole minus 1 plus i at an angle of about 63 degrees and crosses the imaginary axis for K equal to 34.

Exam level

Exercise 6. Stabilising an unstable process

The process G(s) = \dfrac{s + 1}{s(s - 2)} is controlled with a gain K in unity feedback. For which K is the closed loop stable? For which K are the poles real and coincident?

Result: stable for K \gt  2; coincident poles for K = 4 + 2\sqrt 3 = 7.46, at s = -2.73 (and for K = 4 - 2\sqrt 3 = 0.54, at +0.73, unstable).

Solution

Characteristic equation: s^2 + (K - 2)s + K = 0. Stable if all coefficients are positive: K \gt  2.

Coincident poles: zero discriminant, (K - 2)^2 = 4K, K^2 - 8K + 4 = 0, K = 4 \pm 2\sqrt 3. For K = 7.46 the double root is -\frac{K - 2}{2} = -2.73: the breakaway on the left, on the stretch to the left of the zero. For K = 0.54 the root is +0.73, between the two unstable poles: a breakaway in the right half-plane.

The locus is a circle centred at -1 with radius \sqrt 3: -1 \pm\sqrt 3 are exactly 0.73 and -2.73. ✓

Locus of the unstable process
The branches start in the right half-plane, go round the zero at minus 1 and become stable for K greater than 2.

Exercise 7. A stabilising zero

To the system \dfrac{K}{s(s + 2)(s + 4)}, unstable for K \gt  48, a zero at -2.5 is added: L = \dfrac{K(s + 2.5)}{s(s + 2)(s + 4)}. Show that the closed loop is now stable for every K \gt  0.

Result: vertical asymptotes with centroid -1.75; Routh always satisfied.

Solution

Asymptotes: n - m = 2, angles \pm 90°, centroid \frac{(0 - 2 - 4) - (-2.5)}{2} = -1.75. The branches going to infinity stay on the line \text{Re}\,s = -1.75, in the left half-plane.

Routh on s^3 + 6s^2 + (8 + K)s + 2.5K: all coefficients positive, and the third-order condition 6(8 + K) \gt  2.5K holds for every K \gt  0 (it becomes 48 + 3.5K \gt  0).

The zero is obtained with a PD controller, C = K(s + 2.5), or in practice with a lead network (zero followed by a pole much further to the left, which leaves the picture almost unchanged).

Locus with the added zero
The pole at minus 4 goes towards the zero at minus 2.5; the other two branches meet and rise along the vertical line minus 1.75, never crossing the imaginary axis.

Exercise 8. The locus with respect to damping

The mass-spring-damper system \ddot y + a\dot y + 4y = u has adjustable damping a. Sketch the path of the poles as a \ge 0 varies and find a for critical damping and for \zeta = 0.5.

Result: the poles move on the circle |s| = 2; critical damping for a = 4; \zeta = 0.5 for a = 2.

Solution

Characteristic equation s^2 + as + 4 = 0, rewritten as 1 + a\frac{s}{s^2 + 4} = 0: a locus with poles \pm 2i and a zero at 0 (in-depth study).

As long as the poles are complex, their product is 4, so |s|^2 = 4: they lie on the circle of radius \omega_n = 2. With 2\zeta\omega_n = a: \zeta = a/4. Critical damping \zeta = 1 for a = 4 (poles coincident at -2); \zeta = 0.5 for a = 2, poles -1 \pm i\sqrt 3.

Poles as the damping varies
As a increases the poles go down the circle of radius 2 from plus and minus 2 i to minus 2, where they meet for a equal to 4.

Exercise 9. Design to an overshoot specification

The process G(s) = \dfrac{1}{s(s + 5)} is controlled with a gain K. Find K for a 10 % overshoot and the corresponding settling time. Write the MATLAB and Python commands to check it.

Result: \zeta = 0.591, K = 17.9; t_s \approx 1.6 s.

Solution

From the overshoot: \zeta = \frac{-\ln 0.1}{\sqrt{\pi^2 + \ln^2 0.1}} = \frac{2.303}{3.896} = 0.591.

Characteristic equation s^2 + 5s + K: \omega_n = \sqrt K, 2\zeta\omega_n = 5, so \sqrt K = \frac{5}{2\cdot 0.591} = 4.23 and K = 17.9. Poles -2.5 \pm 3.42i.

2 % settling time: \frac{4}{\zeta\omega_n} = \frac{4}{2.5} = 1.6 s, fixed by the real part -2.5, which a gain alone cannot move. For a faster system a zero is needed (lead network) to bring the locus further left.

Check in MATLAB: G = tf(1,[1 5 0]); stepinfo(feedback(17.9*G,1)). In Python: ct.step_info(ct.feedback(17.9*ct.tf([1],[1,5,0]),1)), which gives an overshoot of 10.0 % and a settling time of 1.42 s: the formula 4/\zeta\omega_n is a conservative estimate, obtained from the exponential envelope.

The gain for a 10 per cent overshoot
The dashed green line corresponds to damping 0.591 and meets the vertical branch of the locus for K equal to 17.9.
Subscribe
Notify of
guest
0 Comments
Oldest
Newest Most Voted
0
Would love your thoughts, please comment.x
()
x