CHAPTER 22 / DESIGN BY ROOTS

Sketching a root locus by hand: the construction rules

Seven rules turn the open-loop poles and zeros into a sketch of every closed-loop pole path, with the key numbers marked on it.

THE SHORT ANSWER

Root-locus construction rules sketch the closed-loop pole paths of 1 + KG(s)H(s) = 0 for K from 0 to ∞ without solving for roots. A point lies on the locus when the angles from the zeros minus the angles from the poles total an odd multiple of 180°, and the magnitude condition K = 1/|G(s)H(s)| gives its gain. Real-axis segments, asymptotes, breakaway points, imaginary-axis crossings and departure angles all follow from these two conditions.

∠G(s)H(s)=±180∘(2k+1),K=1∣G(s)H(s)∣\angle G(s)H(s)=\pm180^\circ(2k+1),\qquad K=\frac{1}{|G(s)H(s)|}
WHAT YOU WILL LEARN

Sketch the root locus of a given G(s)H(s) by hand, compute its asymptotes, breakaway points, imaginary-axis crossing and departure angles, and find the gain that gives a specified damping ratio.

The previous chapter traced a root locus by solving a quadratic for each K. A cubic or quartic cannot be handled that way on paper, so we need rules that give the sketch directly from the open-loop poles and zeros.

Review the previous lesson: Root locus: follow the poles
START WITH AN EXAMPLE

First, picture it

Think of the s-plane as a map and the open-loop poles and zeros as landmarks on it. Whether a point s is on the root locus depends only on the directions from the landmarks to that point: the angles from the zeros minus the angles from the poles must add up to an odd multiple of 180°. So you never solve the characteristic equation over and over for different values of K. You compute a few features instead: which stretches of the real axis belong to the locus, in which directions the far branches run, where two poles meet and split apart, and where a branch crosses the imaginary axis. Five or six such features fix the shape of the sketch, and one multiplication gives the gain at any point you care about.

THE IDEA

Build the idea step by step

Start from a negative-feedback loop whose open-loop transfer function is KG(s)H(s) with K > 0. The closed-loop poles satisfy 1 + KG(s)H(s) = 0, that is, KG(s)H(s) = −1. A complex number equals −1 only when its angle is an odd multiple of 180° and its magnitude is 1. The angle condition does not contain K, so it alone decides which points belong to the locus. The magnitude condition then gives the K that puts a closed-loop pole at such a point. Every construction rule below is the angle condition applied in a special place: on the real axis, very far from the origin, right next to a complex pole, or on the imaginary axis. Apply the rules in order, mark each result on the s-plane, and join the marks with smooth curves that are symmetric about the real axis.

Read the equations

1+KG(s)H(s)=0,G(s)H(s)=N(s)D(s)=(s−z1)⋯(s−zm)(s−p1)⋯(s−pn)1+KG(s)H(s)=0,\qquad G(s)H(s)=\frac{N(s)}{D(s)}=\frac{(s-z_1)\cdots(s-z_m)}{(s-p_1)\cdots(s-p_n)}

Characteristic equation. Pull the gain K out in front and write every factor as (s − z) or (s − p). The closed-loop poles are the roots of this equation, one set for each K > 0.

∠G(s)H(s)=∑j=1m∠(s−zj)−∑i=1n∠(s−pi)=±180∘(2k+1),k=0,1,2,…\angle G(s)H(s)=\sum_{j=1}^{m}\angle(s-z_j)-\sum_{i=1}^{n}\angle(s-p_i)=\pm180^\circ(2k+1),\quad k=0,1,2,\dots

Angle condition. A point s is on the locus if and only if the angles from the zeros minus the angles from the poles total an odd multiple of 180°. K does not appear, so the shape of the locus depends only on where the poles and zeros are.

K=1∣G(s)H(s)∣=∏i=1n∣s−pi∣∏j=1m∣s−zj∣K=\frac{1}{|G(s)H(s)|}=\frac{\prod_{i=1}^{n}|s-p_i|}{\prod_{j=1}^{m}|s-z_j|}

Magnitude condition. At a point already known to be on the locus, K is the product of the distances from the poles divided by the product of the distances from the zeros. With no zeros the denominator is 1.

K=0: s=p1,…,pn⟶K→∞: s→z1,…,zm, ∞K=0:\ s=p_1,\dots,p_n\qquad\longrightarrow\qquad K\to\infty:\ s\to z_1,\dots,z_m,\ \infty

Rule 1: branches. The locus has n branches, one from each open-loop pole at K = 0. As K grows, m branches end at the zeros and the other n − m go to infinity. Complex roots come in conjugate pairs, so the plot is symmetric about the real axis.

Nr(σ)∈{1,3,5,… }N_r(\sigma)\in\{1,3,5,\dots\}

Rule 2: real axis. A real point σ is on the locus when N_r(σ), the number of real open-loop poles and zeros to its right, is odd. Complex pairs are not counted.

θk=±180∘(2k+1)n−m,σa=∑pi−∑zjn−m\theta_k=\frac{\pm180^\circ(2k+1)}{n-m},\qquad \sigma_a=\frac{\sum p_i-\sum z_j}{n-m}

Rule 3: asymptotes. The n − m branches that go to infinity approach straight lines that leave the real axis at the centroid σa with angles θk. For n − m = 1 the angle is 180°; for 2 it is ±90°; for 3 it is ±60° and 180°; for 4 it is ±45° and ±135°.

K=−D(s)N(s),dKds=0  ⟺  D′(s)N(s)−D(s)N′(s)=0K=-\frac{D(s)}{N(s)},\qquad \frac{dK}{ds}=0\iff D'(s)N(s)-D(s)N'(s)=0

Rule 4: breakaway and break-in points. Solve dK/ds = 0 and keep only the roots that lie on the locus; for a real root this means K is positive there. Two branches that meet on the real axis leave it at ±90°.

D(jω)+KN(jω)=0 ⟹ Re⁡{⋅}=0,Im⁡{⋅}=0D(j\omega)+KN(j\omega)=0\ \Longrightarrow\ \operatorname{Re}\{\cdot\}=0,\quad \operatorname{Im}\{\cdot\}=0

Rule 5: imaginary-axis crossing. Substitute s = jω in the characteristic polynomial and set its real and imaginary parts to zero; the two equations give the crossing frequency ω and the gain K. The Routh array gives the same result: find the K that makes a row zero, then solve the auxiliary equation formed from the row above.

θd=180∘−∑i≠k∠(pk−pi)+∑j∠(pk−zj)θa=180∘−∑j≠k∠(zk−zj)+∑i∠(zk−pi)\begin{aligned}\theta_d&=180^\circ-\sum_{i\ne k}\angle(p_k-p_i)+\sum_{j}\angle(p_k-z_j)\\\theta_a&=180^\circ-\sum_{j\ne k}\angle(z_k-z_j)+\sum_{i}\angle(z_k-p_i)\end{aligned}

Rule 6: angles of departure and arrival. The departure angle from a complex pole p_k is 180° minus the angles of the vectors drawn to p_k from the other poles, plus the angles of the vectors from the zeros. The arrival angle at a complex zero z_k swaps the roles of poles and zeros.

s=−ζωn±jωn1−ζ2,cos⁡β=ζ,K=1∣G(s)H(s)∣s=-\zeta\omega_n\pm j\omega_n\sqrt{1-\zeta^2},\qquad \cos\beta=\zeta,\qquad K=\frac{1}{|G(s)H(s)|}

Rule 7: gain at a chosen point. Poles with damping ratio ζ lie on the rays that make an angle β = cos⁻¹ζ with the negative real axis. Find the point on that ray that satisfies the angle condition, then apply the magnitude condition there to get K.

What the symbols mean

KK
adjustable loop gain, K > 0; its unit is whatever makes KG(s)H(s) dimensionless
G(s)H(s)G(s)H(s)
open-loop transfer function with the gain K taken out
N(s), D(s)N(s),\,D(s)
numerator and denominator polynomials of G(s)H(s), each with leading coefficient 1
ss
complex frequency, rad/s (1/s)
pi, zjp_i,\,z_j
open-loop poles and zeros, rad/s
n, mn,\,m
number of open-loop poles and zeros, with n ≥ m
Nr(σ)N_r(\sigma)
number of real open-loop poles and zeros to the right of the real point σ
θk\theta_k
angle of an asymptote from the positive real direction, degrees
σa\sigma_a
centroid: the real-axis point where the asymptotes meet, rad/s
θd, θa\theta_d,\,\theta_a
angle of departure from a complex pole and angle of arrival at a complex zero, degrees
ω\omega
frequency at which a branch crosses the imaginary axis, rad/s
ζ\zeta
damping ratio of a complex pole pair, dimensionless
ωn\omega_n
undamped natural frequency: the distance of the pole pair from the origin, rad/s
β\beta
angle between the negative real axis and the ray to the pole, degrees

These rules are for negative feedback with K running from 0 to +∞, and for G(s)H(s) written with factors (s − z) and (s − p) so that K is the only constant in front. For positive feedback or negative K the angle condition becomes an even multiple of 180°, and the real-axis rule and the asymptote angles change with it.

WORKED EXAMPLE

Sketch the locus of K/(s(s+1)(s+2)) and choose K for ζ = 0.5

  1. Set up. With unity feedback the characteristic equation is a cubic. There are n = 3 poles at 0, −1 and −2 and no zeros, so rule 1 gives three branches that all go to infinity.

    G(s)H(s)=1s(s+1)(s+2),s3+3s2+2s+K=0G(s)H(s)=\frac{1}{s(s+1)(s+2)},\qquad s^3+3s^2+2s+K=0
  2. Rule 2. Between −1 and 0 there is one pole to the right, and to the left of −2 there are three. Those two stretches of the real axis are on the locus; the stretch between −2 and −1 is not.

    −1≤σ≤0,σ≤−2-1\le\sigma\le0,\qquad \sigma\le-2
  3. Rule 3. With n − m = 3 the asymptotes are at ±60° and 180°, and they meet at the centroid −1.

    θk=180∘(2k+1)3=60∘, 180∘, −60∘,σa=0−1−23=−1\theta_k=\frac{180^\circ(2k+1)}{3}=60^\circ,\ 180^\circ,\ -60^\circ,\qquad \sigma_a=\frac{0-1-2}{3}=-1
  4. Rule 4. Solve dK/ds = 0. The root −0.423 lies on the segment from −1 to 0 and gives K = 0.385, so it is the breakaway point. The root −1.577 lies between −2 and −1, off the locus, and gives K = −0.385, so it is rejected.

    K=−(s3+3s2+2s),dKds=−(3s2+6s+2)=0 ⇒ s=−1±13=−0.423, −1.577K=-(s^3+3s^2+2s),\quad \frac{dK}{ds}=-(3s^2+6s+2)=0\ \Rightarrow\ s=-1\pm\frac{1}{\sqrt3}=-0.423,\ -1.577
  5. Rule 5. Substitute s = jω. The imaginary part gives ω = √2 rad/s and the real part then gives K = 6. The Routh array agrees: its s¹ entry (6 − K)/3 is zero at K = 6, and the auxiliary equation 3s² + 6 = 0 gives s = ±j√2.

    (K−3ω2)+jω(2−ω2)=0 ⇒ ω=2≈1.414,K=6(K-3\omega^2)+j\omega(2-\omega^2)=0\ \Rightarrow\ \omega=\sqrt2\approx1.414,\quad K=6
  6. Rule 6 and the sketch. There are no complex open-loop poles, so there is no departure angle to compute. The branches from 0 and −1 meet at −0.423, leave the real axis at ±90°, bend toward the ±60° asymptotes and cross the imaginary axis at ±j1.414. The branch from −2 moves left along the real axis.

    s1,2=−0.423 (K=0.385) ⟶ s1,2=±j1.414 (K=6)s_{1,2}=-0.423\ (K=0.385)\ \longrightarrow\ s_{1,2}=\pm j1.414\ (K=6)
  7. Rule 7, the ray. For ζ = 0.5 the angle is β = 60°, so the test point is s = r∠120°. Then s³ = r³ is real, and the imaginary part of the characteristic equation fixes r.

    s=r(−12+j32):Im⁡: −332r2+3 r=0 ⇒ r=23s=r\left(-\tfrac12+j\tfrac{\sqrt3}{2}\right):\quad \operatorname{Im}:\ -\tfrac{3\sqrt3}{2}r^2+\sqrt3\,r=0\ \Rightarrow\ r=\tfrac23
  8. Rule 7, the gain. The real part gives K. The dominant poles are at −0.333 ± j0.577 and the gain is 28/27 ≈ 1.04.

    Re⁡: r3−32r2−r+K=0 ⇒ K=2827≈1.04,s=−13±j13\operatorname{Re}:\ r^3-\tfrac32r^2-r+K=0\ \Rightarrow\ K=\tfrac{28}{27}\approx1.04,\qquad s=-\tfrac13\pm j\tfrac{1}{\sqrt3}
  9. Check with the two conditions. The angles from the three poles add to 180°, and the product of the three distances gives the same K. The poles sum to −3, so the third pole is at −3 + 2/3 = −2.33, far enough left for the pair to dominate.

    120∘+40.9∘+19.1∘=180∘,K=∣s∣ ∣s+1∣ ∣s+2∣=0.667×0.882×1.764=1.04120^\circ+40.9^\circ+19.1^\circ=180^\circ,\qquad K=|s|\,|s+1|\,|s+2|=0.667\times0.882\times1.764=1.04
REFERENCE

Reference table

Root-locus construction rules (negative feedback, K from 0 to +∞)
RuleWhat it givesFormulaExample: K/[s(s + 1)(s + 2)]
1. Branchesn branches start at the open-loop poles; m end at the zeros and n − m go to infinityK=0: s=pi,K→∞: s→zj, ∞K=0:\ s=p_i,\quad K\to\infty:\ s\to z_j,\ \infty3 branches from 0, −1 and −2, all going to infinity
2. Real axisreal points with an odd number of real poles and zeros to their rightNr(σ)∈{1,3,5,… }N_r(\sigma)\in\{1,3,5,\dots\}−1 ≤ σ ≤ 0 and σ ≤ −2
3. Asymptotesdirections and centroid of the branches that go to infinityθk=±180∘(2k+1)n−m,σa=∑pi−∑zjn−m\theta_k=\dfrac{\pm180^\circ(2k+1)}{n-m},\quad \sigma_a=\dfrac{\sum p_i-\sum z_j}{n-m}±60° and 180°, centroid σa = −1
4. Breakaway and break-inwhere branches leave or join the real axisdKds=0,K>0\dfrac{dK}{ds}=0,\quad K>0s = −0.423 with K = 0.385; the root −1.577 gives K = −0.385 and is rejected
5. Imaginary-axis crossingthe gain and frequency at the stability boundaryD(jω)+KN(jω)=0D(j\omega)+KN(j\omega)=0s = ±j√2 ≈ ±j1.414 at K = 6
6. Departure and arrival anglesthe direction in which a branch leaves a complex pole or reaches a complex zeroθd=180∘−∑i≠k∠(pk−pi)+∑j∠(pk−zj)\theta_d=180^\circ-\sum_{i\ne k}\angle(p_k-p_i)+\sum_{j}\angle(p_k-z_j)not needed: no complex poles or zeros
7. Gain at a chosen pointthe K that puts a closed-loop pole at a point on the locusK=1∣G(s)H(s)∣K=\dfrac{1}{|G(s)H(s)|}ζ = 0.5: s = −0.333 ± j0.577 at K ≈ 1.04; the third pole is at −2.333

G(s)H(s) is written with factors (s − z) and (s − p) so that K is the only constant in front. For positive feedback or negative K the angle condition becomes an even multiple of 180°.

More detail and model assumptions
A CLOSER LOOK

The angle condition as geometry

Write G(s)H(s) as a product of factors (s − z) over a product of factors (s − p). For a test point s, the factor (s − p) is the vector drawn from p to s. Its angle is measured counterclockwise from the positive real direction, and its magnitude is the distance from p to s. The angle of G(s)H(s) is therefore the sum of the angles from the zeros minus the sum of the angles from the poles, and its magnitude is the product of the distances from the zeros divided by the product of the distances from the poles. Both conditions can be checked on a sketch with a protractor and a ruler.

The real-axis rule follows at once. Put the test point on the real axis. A complex-conjugate pair contributes two angles that cancel, +θ and −θ. A real pole or zero to the left of the test point contributes 0°, and one to the right contributes 180°. The total is an odd multiple of 180° exactly when the number of real poles and zeros to the right is odd.

The asymptote rule comes from looking at the pole-zero pattern from far away. From a distant point all n + m vectors are nearly parallel with a common angle θ, so the angle condition reads (m − n)θ = odd multiple of 180°, which gives θ = (2k+1)180°/(n − m). To find where these lines meet, divide the polynomials: D(s)/N(s) = s^(n−m) − (Σp − Σz)s^(n−m−1) + …, and this matches (s − σa)^(n−m) to the first two terms when σa = (Σp − Σz)/(n − m). Imaginary parts cancel in the sums, so the centroid is always a real number.

A CLOSER LOOK

Why dK/ds = 0 marks a breakaway point

On a real-axis segment of the locus, K = −D(s)/N(s) is a real function of the real variable s. Between two adjacent poles K is zero at both ends and positive in between, so it has a maximum somewhere inside the segment. Below that maximum each value of K is reached at two real points, which are two real closed-loop poles. At the maximum the two poles coincide, and above it they can only continue as a complex pair. That is why the breakaway point is where dK/ds = 0. Between two zeros K is infinite at both ends and has a minimum, which marks a break-in point.

The same condition says that the characteristic polynomial D(s) + KN(s) has a repeated root. A repeated root makes both the polynomial and its derivative zero, and eliminating K from those two equations gives D′(s)N(s) − D(s)N′(s) = 0. When two branches meet on the real axis they arrive and leave at right angles to each other, so a pair breaking away from the real axis leaves at ±90°.

The equation dK/ds = 0 usually has more roots than there are breakaway points. Test every root: a real root is valid if it lies on a real-axis segment of the locus, and the K computed there is then positive. A complex root is valid only if it satisfies the angle condition, which happens rarely.

A CLOSER LOOK

Using the sketch for design

A damping-ratio specification is a ray on the s-plane: poles with damping ratio ζ lie at an angle β = cos⁻¹ζ from the negative real axis. Where the ray meets the locus you get the dominant poles, and the magnitude condition gives the gain. The overshoot predicted from ζ is accurate only when that pair dominates, which needs the remaining poles to be several times farther to the left. In the worked example the third pole is at −2.33, seven times farther left than the pair at −0.333, so the second-order estimate is good.

The K in the magnitude condition multiplies the form with factors (s − p). If a problem gives the loop in time-constant form, convert it first: K_B/(s(1+s)(1+0.5s)) equals 2K_B/(s(s+1)(s+2)), so the root-locus gain is K = 2K_B. Mixing the two forms is the most common way to lose a factor in an exam answer.

The imaginary-axis crossing also measures how much gain is in reserve. If the loop runs at K = 1.04 and the crossing is at K = 6, the gain can rise by a factor of 6/1.04 ≈ 5.8 before the loop becomes unstable. The Bode chapters call this ratio the gain margin, and they find the same crossing frequency √2 rad/s as the phase crossover frequency.

INTERACTIVE / 1 + K·G(s) = 0Pick or drag the poles and zeros, then move K
−6−4−22−4j−2j2j4jK = 2.00jωσ
  • open-loop pole
  • open-loop zero
  • closed-loop pole at K
  • breakaway / break-in
  • imaginary-axis crossing
  • asymptotes, ◆ centroid
Up to 6 poles and 4 zeros, never more zeros than poles. Select a × or ○ to move it.
Branches
3
Asymptotes
3
σa
−1.00
jω-crossing K
6.00
Closed loop stable

3 open-loop poles and 0 zeros: 3 branches start at the poles (×) when K = 0; there are no zeros, so 3 asymptotes at −60°, 60°, 180° meet at σa = −1.00. Breakaway/break-in: s = −0.42 (K = 0.38). Branches reach the imaginary axis at K = 6.00 (ω = ±1.41 rad/s). At K = 2.00 the closed-loop poles are −0.24 ± 0.86j, −2.52: stable; the dominant complex pair has ζ ≈ 0.27.

Interactive root-locus explorer: pick or drag the open-loop poles and zeros, move K and follow where the closed-loop poles go. Try it online at eigenroom.com/theory/root-locus-rules.

TRY IT

Connect the example to the lab

The root-locus explorer in the previous chapter (/theory/root-locus) draws exactly the plant of the worked example, G(s) = 1/(s(s+1)(s+2)). Set K = 0.35 and then K = 0.40: the two right-hand poles meet near −0.42 and leave the real axis, as the breakaway calculation says. Set K = 1.05: the explorer reports a damping ratio of about 0.50 with poles near −0.33 ± 0.58j. Set K = 6: the pair sits on the imaginary axis at ±1.41j. For the Routh side of rule 5, open the Routh bench below and enter a = 3, b = 2, c = K, because the characteristic polynomial s³ + 3s² + 2s + K is the bench's s³ + as² + bs + c.

Check the axis crossing on the Routh bench
YOUR EXPERIMENT

Make a prediction before moving a control.

  1. PREDICTThe worked example says s³ + 3s² + 2s + K is stable for 0 < K < 6 and has poles at ±j√2 when K = 6. Before touching the bench, write the first column of the Routh array and the number of right-half-plane roots for K = 5, 6 and 7.
  2. CHANGE ONE THINGOn the Routh bench set a = 3 and b = 2, then change only c, which plays the role of K, through 5, 6 and 7.
  3. OBSERVERead the s¹ entry (6 − c)/3, the stability status and the list of poles. At c = 6 the bench reports a zero row and the factorization (s + 3)(s² + 2).
  4. EXPLAINExplain why the zero row gives the crossing frequency, and compute that frequency from the auxiliary equation 3s² + 6 = 0. Then say which of the three poles at c = 7 have moved into the right half-plane.
BE CAREFUL

A root of dK/ds = 0 is only a candidate for a breakaway point. Keep it only if it lies on the locus, which for a real root means the K computed there is positive. In the worked example s = −1.577 solves dK/ds = 0 but gives K = −0.385, so it is not a breakaway point for K > 0.

Check your understanding+

The locus of K/(s(s+1)(s+2)) has asymptotes at ±60° and 180° from the centroid −1. Without finding any roots, what does this tell you about large K?

Two branches follow the ±60° lines into the right half-plane, so the closed loop must become unstable above some finite gain. Rule 5 finds that gain: K = 6, where the branches cross the imaginary axis at ω = √2 rad/s.
PRACTICE PROBLEMS

Solve these before opening the solutions

Work each problem on paper first. Open the answer to check yours, then the full solution if you need the steps, and mark how it went.

  1. For the open loop below, state the number of branches, the real-axis segments of the locus, and the angles and centroid of the asymptotes.

    G(s)H(s)=K(s+1)s(s+2)(s+4)(s+6)G(s)H(s)=\frac{K(s+1)}{s(s+2)(s+4)(s+6)}
    Show answer

    AnswerFour branches. The real-axis segments are from −1 to 0, from −4 to −2, and everything to the left of −6. Three asymptotes at ±60° and 180° leave from the centroid −11/3 ≈ −3.67.

    θk=±60∘, 180∘,σa=−113≈−3.67\theta_k=\pm60^\circ,\ 180^\circ,\qquad \sigma_a=-\tfrac{11}{3}\approx-3.67
    Show full solution

    Solution

    1. Count poles and zeros. There are four poles and one zero, so there are four branches. One ends at the zero and three go to infinity.

      n=4,m=1,n−m=3n=4,\quad m=1,\quad n-m=3
    2. Apply the real-axis rule segment by segment. The count of real poles and zeros to the right is 1 between −1 and 0, 2 between −2 and −1, 3 between −4 and −2, 4 between −6 and −4, and 5 to the left of −6. The odd counts are on the locus.

      −1≤σ≤0,−4≤σ≤−2,σ≤−6-1\le\sigma\le0,\qquad -4\le\sigma\le-2,\qquad \sigma\le-6
    3. Compute the asymptotes. Subtract the sum of the zeros from the sum of the poles and divide by n − m.

      σa=(0−2−4−6)−(−1)3=−113,θk=180∘(2k+1)3\sigma_a=\frac{(0-2-4-6)-(-1)}{3}=-\frac{11}{3},\qquad \theta_k=\frac{180^\circ(2k+1)}{3}
  2. Find the breakaway point, the break-in point and the gain at each. Then show that the part of the locus off the real axis is a circle.

    G(s)H(s)=K(s+3)s(s+1)G(s)H(s)=\frac{K(s+3)}{s(s+1)}
    Show answer

    AnswerBreakaway at s = −3 + √6 ≈ −0.551 with K ≈ 0.101; break-in at s = −3 − √6 ≈ −5.449 with K ≈ 9.90. Between these gains the poles move on a circle centred at the zero −3 with radius √6 ≈ 2.45.

    s=−3±6,K=5∓26=0.101, 9.90s=-3\pm\sqrt6,\qquad K=5\mp2\sqrt6=0.101,\ 9.90
    Show full solution

    Solution

    1. Find the real-axis segments first: from −1 to 0, and to the left of −3. Two poles start on the first segment, so they must break away from it; one branch must end at the zero and one at −∞, so the pair must come back to the real axis left of −3.

      −1≤σ≤0,σ≤−3-1\le\sigma\le0,\qquad \sigma\le-3
    2. Write K as a function of s and set its derivative to zero.

      K=−s(s+1)s+3,(2s+1)(s+3)−(s2+s)=s2+6s+3=0 ⇒ s=−3±6K=-\frac{s(s+1)}{s+3},\qquad (2s+1)(s+3)-(s^2+s)=s^2+6s+3=0\ \Rightarrow\ s=-3\pm\sqrt6
    3. Both roots lie on real-axis segments of the locus. Evaluate K at each.

      K(−0.551)=0.551×0.4492.449=0.101,K(−5.449)=5.449×4.4492.449=9.90K(-0.551)=\frac{0.551\times0.449}{2.449}=0.101,\qquad K(-5.449)=\frac{5.449\times4.449}{2.449}=9.90
    4. For the circle, use the characteristic equation s² + (1+K)s + 3K = 0. For complex roots σ ± jω the sum of the roots gives 2σ = −(1+K) and the product gives σ² + ω² = 3K. Eliminate K.

      σ2+ω2=−3(2σ+1) ⇒ (σ+3)2+ω2=6\sigma^2+\omega^2=-3(2\sigma+1)\ \Rightarrow\ (\sigma+3)^2+\omega^2=6
  3. Find the asymptotes, the breakaway points, the range of K for which the closed loop is stable, and the frequency of oscillation at the stability limit.

    G(s)H(s)=Ks(s+2)(s+4)(s+6)G(s)H(s)=\frac{K}{s(s+2)(s+4)(s+6)}
    Show answer

    AnswerAsymptotes at ±45° and ±135° from the centroid −3. Breakaway points at s ≈ −0.764 and s ≈ −5.236, both at K = 16. The closed loop is stable for 0 < K < 160 and oscillates at 2 rad/s when K = 160.

    σa=−3,s=−3±5 (K=16),0<K<160,ω=2 rad/s\sigma_a=-3,\quad s=-3\pm\sqrt5\ (K=16),\quad 0<K<160,\quad \omega=2\ \mathrm{rad/s}
    Show full solution

    Solution

    1. Asymptotes and real axis. With four poles and no zeros there are four asymptotes. The real-axis segments are from −2 to 0 and from −6 to −4.

      σa=0−2−4−64=−3,θk=180∘(2k+1)4=±45∘, ±135∘\sigma_a=\frac{0-2-4-6}{4}=-3,\qquad \theta_k=\frac{180^\circ(2k+1)}{4}=\pm45^\circ,\ \pm135^\circ
    2. Breakaway points. Group the factors as (s² + 6s)(s² + 6s + 8) and let u = s² + 6s. Then dK/ds = 0 when s = −3 or u = −4.

      K=−u(u+8),dKds=−(2u+8)(2s+6)=0K=-u(u+8),\qquad \frac{dK}{ds}=-(2u+8)(2s+6)=0
    3. Test the candidates. u = −4 gives s = −3 ± √5, both on the locus, with K = 16. The root s = −3 lies between −4 and −2, off the locus, and gives K = −9, so it is rejected.

      s2+6s+4=0 ⇒ s=−0.764, −5.236,K=−(−4)(4)=16s^2+6s+4=0\ \Rightarrow\ s=-0.764,\ -5.236,\qquad K=-(-4)(4)=16
    4. Routh array for s⁴ + 12s³ + 44s² + 48s + K. The first column is positive when 0 < K < 160.

      s4144Ks31248s240Ks148−0.3Ks0K\begin{array}{c|ccc}s^4&1&44&K\\s^3&12&48&\\s^2&40&K&\\s^1&48-0.3K&&\\s^0&K&&\end{array}
    5. At K = 160 the s¹ row is zero. The auxiliary equation from the s² row gives the crossing points.

      40s2+160=0 ⇒ s=±j240s^2+160=0\ \Rightarrow\ s=\pm j2
  4. Find the angle of departure from the pole at −3 + j2, show that there is no breakaway point, and find where the locus crosses the imaginary axis.

    G(s)H(s)=Ks(s2+6s+13)G(s)H(s)=\frac{K}{s(s^2+6s+13)}
    Show answer

    AnswerThe departure angle is about −56.3° from −3 + j2 and +56.3° from its conjugate. dK/ds = 0 has no real root, so no branches meet on the real axis. The locus crosses the imaginary axis at ω = √13 ≈ 3.61 rad/s when K = 78.

    θd=−56.3∘,ω=13≈3.61 rad/s,K=78\theta_d=-56.3^\circ,\qquad \omega=\sqrt{13}\approx3.61\ \mathrm{rad/s},\quad K=78
    Show full solution

    Solution

    1. Locate the features that need no calculation. The poles are at 0 and −3 ± j2. The whole negative real axis is on the locus, and there are three asymptotes.

      σa=0−3−33=−2,θk=±60∘, 180∘\sigma_a=\frac{0-3-3}{3}=-2,\qquad \theta_k=\pm60^\circ,\ 180^\circ
    2. Departure angle. The vector from the pole at 0 to −3 + j2 has angle 180° − tan⁻¹(2/3) = 146.3°. The vector from the conjugate pole points straight up, 90°.

      θd=180∘−146.3∘−90∘=−56.3∘\theta_d=180^\circ-146.3^\circ-90^\circ=-56.3^\circ
    3. Breakaway test. The derivative of K is a quadratic with a negative discriminant, so it has no real root.

      dKds=−(3s2+12s+13),122−4⋅3⋅13=−12<0\frac{dK}{ds}=-(3s^2+12s+13),\qquad 12^2-4\cdot3\cdot13=-12<0
    4. Imaginary-axis crossing. Substitute s = jω in s³ + 6s² + 13s + K and separate the parts.

      (K−6ω2)+jω(13−ω2)=0 ⇒ ω2=13,K=78(K-6\omega^2)+j\omega(13-\omega^2)=0\ \Rightarrow\ \omega^2=13,\quad K=78
  5. Unity feedback. Compute the asymptotes, the breakaway point and the imaginary-axis crossing. Then find K so that the dominant closed-loop poles have ζ = 1/√2 ≈ 0.707, locate the third pole, and state by what factor K can still be raised before the loop becomes unstable.

    G(s)=Ks(s+2)(s+6)G(s)=\frac{K}{s(s+2)(s+6)}
    Show answer

    AnswerAsymptotes at ±60° and 180° from −8/3; breakaway at s ≈ −0.903 (K ≈ 5.05); crossing at ω = √12 ≈ 3.46 rad/s with K = 96. For ζ = 0.707 the gain is K ≈ 8.88, with poles at −0.838 ± j0.838 and a third pole at −6.32. The gain can still be raised by a factor of 96/8.88 ≈ 10.8.

    K≈8.88,s=−0.838±j0.838, −6.32,968.88≈10.8K\approx8.88,\qquad s=-0.838\pm j0.838,\ -6.32,\qquad \frac{96}{8.88}\approx10.8
    Show full solution

    Solution

    1. Asymptotes and real axis. The real-axis segments are from −2 to 0 and to the left of −6.

      σa=0−2−63=−83,θk=±60∘, 180∘\sigma_a=\frac{0-2-6}{3}=-\frac83,\qquad \theta_k=\pm60^\circ,\ 180^\circ
    2. Breakaway point. Of the two roots only −0.903 lies on the locus.

      K=−(s3+8s2+12s),3s2+16s+12=0 ⇒ s=−0.903, −4.431;K(−0.903)=5.05K=-(s^3+8s^2+12s),\quad 3s^2+16s+12=0\ \Rightarrow\ s=-0.903,\ -4.431;\qquad K(-0.903)=5.05
    3. Imaginary-axis crossing from s = jω.

      (K−8ω2)+jω(12−ω2)=0 ⇒ ω=12≈3.46,K=96(K-8\omega^2)+j\omega(12-\omega^2)=0\ \Rightarrow\ \omega=\sqrt{12}\approx3.46,\quad K=96
    4. The ζ = 0.707 ray is at 135°. Put s = r∠135°, so that s² = −jr² and s³ = r³∠45°. The imaginary part gives a quadratic in r; take the smaller root, because the larger one, r = 10.13, gives a negative K.

      Im⁡: r2−82 r+12=0 ⇒ r=1.185;Re⁡: K=12r−r32=8.88\operatorname{Im}:\ r^2-8\sqrt2\,r+12=0\ \Rightarrow\ r=1.185;\qquad \operatorname{Re}:\ K=\frac{12r-r^3}{\sqrt2}=8.88
    5. Poles and margin. The pair is at −r/√2 ± jr/√2. The poles sum to −8, which gives the third pole; it is 7.5 times farther left than the pair, so the pair dominates.

      s1,2=−0.838±j0.838,s3=−8+2(0.838)=−6.32,968.88=10.8s_{1,2}=-0.838\pm j0.838,\qquad s_3=-8+2(0.838)=-6.32,\qquad \frac{96}{8.88}=10.8

Mark each exercise after you check your answer. Marks stay in this browser.

KEY TERMS

Terms this lecture teaches

All terms in the glossary →
QUESTIONS STUDENTS ASK

Frequently asked questions

How do I find the asymptotes of a root locus?

With n open-loop poles and m zeros, n − m branches go to infinity along lines at angles ±180°(2k + 1)/(n − m) that meet on the real axis at σa = (Σp − Σz)/(n − m). For K/(s(s + 1)(s + 2)), n − m = 3, so the asymptotes leave the centroid −1 at ±60° and 180°.

Is every root of dK/ds = 0 a breakaway point?

No, each root is only a candidate. Keep a real root only if it lies on a real-axis segment of the locus, which means the gain computed there is positive. For K/(s(s + 1)(s + 2)), s = −0.423 gives K = 0.385 and is the breakaway point, while s = −1.577 gives K = −0.385 and is rejected.

Which parts of the real axis belong to the root locus?

A real point is on the locus when the number of real open-loop poles and zeros to its right is odd. Complex-conjugate pairs are not counted because their angle contributions cancel. The rule assumes negative feedback with K > 0; for positive feedback the count must be even.

How do I choose K for a desired damping ratio?

Draw the ray at β = cos⁻¹ζ from the negative real axis, find where it meets the locus using the angle condition, then apply the magnitude condition there. For K/(s(s + 1)(s + 2)) and ζ = 0.5, the dominant poles are −0.333 ± j0.577 and K = 28/27 ≈ 1.04, with the third pole at −2.33.

Why does my root-locus gain come out wrong by a constant factor?

The magnitude condition gives the gain in front of factors written as (s − p) and (s − z). A loop given in time-constant form must be converted first: 0.5/(s(1 + s)(1 + 0.5s)) equals 1/(s(s + 1)(s + 2)), so its root-locus gain is K = 1, not 0.5.

Why the next lesson follows

The crossing at K = 6 came from substituting s = jω into the characteristic equation. The next chapter looks at G(jω) for its own sake: it is the response of the system to a sine wave, and it leads to Bode plots, where K = 6 and ω = √2 rad/s reappear as the gain margin and the phase crossover frequency.

YOUR PROGRESS

Finished the reading and the exercises? Mark the lecture as done to tick it on the syllabus. Progress stays in this browser.

EIGENROOM · https://eigenroom.com/theory/root-locus-rules · Updated 2026-11-15