Flow Lab

Syllabus / Module 13: Applications: Control Systems

Lesson 13.4

Stability: Routh–Hurwitz & Root Locus

BIBO stability, Routh–Hurwitz criterion, root-locus sketching rules, and designing gains by moving poles.

Written byClaude Sonnet

Prerequisites:13.3 PID Control
🌊
Why this matters for simulation: Choosing autopilot gains to damp the phugoid.

By the end of this lesson you can…

  • State the pole condition for BIBO and asymptotic stability and apply it to the characteristic polynomial of a closed loop.
  • Build a Routh array, count right-half-plane roots from the first-column sign changes, and handle a zero pivot and a row of zeros.
  • Use Routh to find the stable range of a gain and the frequency at its edge, recovering the PI edge $K_i = 1.1(1+K_p)$ of the cruise loop.
  • Sketch a root locus from the rules: branches, real-axis segments, asymptotes and centroid, breakaway points and imaginary-axis crossings.
  • Read damping off a locus and pick a gain, for the cruise loop and for the yaw damper of lesson 12.6.
  • Approximate a delay by a Pade formula and compare its critical gain with the exact $16.351$.
On this page
·
  1. What stability means
  2. The Routh-Hurwitz test
  3. The root locus
  4. The yaw damper of Module 12
  5. Delay: a transfer function that is not a polynomial
  6. More worked examples
  7. Concept checks
  8. Playground and exercise
  9. Problems
  10. What comes next
  11. Key Takeaways

13.3 PID Control ended on a warning. At Kp=4K_p = 4 the cruise loop with an engine lag was fine at Ki=0.5K_i = 0.5 and unstable at Ki=6K_i = 6, and the edge between them was Ki=1.1(1+Kp)=5.5K_i = 1.1(1 + K_p) = 5.5, found by putting s=iωs = i\omega into one polynomial. A simulator or an autopilot needs the general answers: is this loop stable, and if I turn one knob, where do all the poles go? Routh-Hurwitz answers the first without a root finder, and the root locus draws every pole at once. A delay, at the end, will show where both fall short.

▶ Video Stability as poles in the left half plane, the Routh array built row by row, the cruise loop's integral gain crossing the imaginary axis, and the root locus rules applied to the cruise plant and the yaw dampercaptions only
Transcript
  1. A closed loop is stable when every pole sits in the open left half plane: the standard statement, quoted from memory. Marginal is not safe. The poles that matter are the roots of 1 plus L of s, the characteristic equation D plus K times N. Take the cruise PI loop with K p equal 4. At K i equal 0.5 all three poles are in the green half. At K i equal 6 a pair has crossed into the red, at plus 0.0141 plus or minus 0.7291 i.
  2. The Routh array tests the polynomial itself. Write the coefficients alternately in two rows and build each later row from the two above. The rule, from memory and not proved here: the number of right-half-plane roots equals the number of sign changes in the first column. At K i equal 0.5 the first column is 10, 11, 4.5455, 0.5: no sign change. At K i equal 6 it is 10, 11, minus 0.4545, 6: two sign changes, so two unstable roots, counted without computing them.
  3. Now recover the edge that 13.3 found by substituting s equal i omega. The first column has no sign change exactly when K i is below 1.1 times 1 plus K p: 5.5 at K p equal 4. At the edge the s to the 1 row is all zeros, and the row above gives 11 s squared plus 5.5, so the oscillation is at plus or minus 0.7071 i, with the third root at minus 1.1. With derivative gain the 11 becomes 11 plus K d: at K p equal 10 the edge is 12.1, and 18.7 with K d equal 6.
  4. The root locus plots every closed-loop pole as K runs from 0 to infinity. A point is on it when K equals minus D over N is a positive real number, which gives the angle condition: the angle of N minus the angle of D is 180 degrees. Three rules are from memory: n branches, symmetric about the real axis; they start at the poles and end at zeros or infinity; a real point is on the locus if the count of real poles and zeros to its right is odd. Three are derived: asymptotes through the centroid sigma, breakaway points where D N prime minus D prime N is zero, and crossings at the Routh edge.
  5. Now two cruise loci. Left: proportional gain on the engine lag, poles at minus 0.1 and minus 1. The branches break away at s equal minus 0.55, at K equal 2.025, matching 13.2. The pole sum is fixed, so the pair then climbs the vertical line through minus 0.55. K equal 4 gives damping 0.7778, K equal 19 gives 0.3889, and no gain is unstable. Right: the integral gain at K p equal 4. The centroid is minus 0.3667 and the discriminant is negative, so no breakaway. The complex branches run out at 60 degrees and cross at K i equal 5.5, omega 0.7071.
  6. A locus by hand: K over s times s plus 1 times s plus 2. The centroid is minus 1. Breakaway: 3 s squared plus 6 s plus 2 equals 0, so s equal minus 0.4226 at K equal 0.3849. Below that all roots are real. Above it a pair bends toward the 60 degree lines and crosses where Routh gives K equal 6, auxiliary polynomial 3 s squared plus 6, omega 1.4142. At K equal 1 the damping is 0.5148.
  7. The yaw damper of Module 12. The rudder feedback makes the characteristic polynomial linear in K, so the same rules apply to a fourth-order system. At K equal 0.5 the Dutch-roll pair is at minus 1.6647 plus or minus 1.5465 i, damping 0.7326. The pair breaks away at K equal 0.7165, with best damping 0.9776 at K equal 0.7. The branch moving right meets the spiral root at the break-in, K equal 0.9665, and that pair is bound for the zeros at minus 0.0509 plus or minus 0.5797 i: damping 0.5124 at K equal 2, 0.3711 at 3.
  8. A delay is not rational, so the rules no longer apply. To use polynomial tools, approximate. The exact critical gain from 13.1 is 16.351. The first-order Pade form gives a quadratic stable up to K equal 21.0000, at omega 2.0976: 28 percent above the exact value, an unsafe error. The second-order form gives 16.4866, 0.8 percent above. Pade puts a zero at plus 2, in the right half plane: the signature of a delay. At K equal 14.5 the two forms give damping 0.1846 and 0.0587.
  9. A loop is stable when every pole is in the left half plane, and Routh counts the bad ones from the coefficients alone. It recovered the integral edge of 5.5 at K p equal 4, and the root locus showed the cruise breakaway at minus 0.55 and the yaw damper breakaway at 0.7165. A delay was the gap: Pade puts the critical gain at 21, against the exact 16.351. Lesson 13.5 is not yet written. It is meant to take the exact route, gain and phase against frequency as a Bode plot, with the margin to instability read off it, for a model or a measured system.

What stability means

The standard statements (from memory of the textbook theorems): a transfer function with common factors cancelled is BIBO stable, every bounded input giving a bounded output, exactly when all its poles lie in the open left half plane Re s<0\text{Re}\,s \lt 0; and x˙=Ax\dot x = Ax is asymptotically stable exactly when every eigenvalue of AA has Re λ<0\text{Re}\,\lambda \lt 0 (7.2 Linear Systems & the Eigenvalue Method). A simple pair on the imaginary axis rings forever, and a bounded input at that frequency gives an unbounded output, the resonance of 5.6 Forced Oscillations & Resonance: marginal is not safe.

In a loop the poles that matter are the roots of 1+L(s)=01 + L(s) = 0. Write L=K N(s)/D(s)L = K\,N(s)/D(s); the characteristic equation is

χ(s)=D(s)+K N(s)=0.\chi(s) = D(s) + K\,N(s) = 0 .

For the PID loop of 13.3 it was 10s3+(11+Kd)s2+(1+Kp)s+Ki10s^3 + (11 + K_d)s^2 + (1+K_p)s + K_i. Test the whole loop, not a cancelled TT: a pole cancelled against a zero is hidden from TT but not from the plant.

The Routh-Hurwitz test

A polynomial with an>0a_n \gt 0 whose roots all have negative real part has every coefficient positive (products of (s+c)(s+c) and (s2+bs+c)(s^2+bs+c) factors). A missing or negative coefficient proves instability; the converse holds only up to degree two.

The array. Write the coefficients alternately in two rows and build each new row from the two above:

snanan−2⋯sn−1an−1an−3⋯sn−2b1b2⋯b1=an−1an−2−anan−3an−1,b2=an−1an−4−anan−5an−1.\begin{array}{c|ccc} s^n & a_n & a_{n-2} & \cdots \\ s^{n-1} & a_{n-1} & a_{n-3} & \cdots \\ s^{n-2} & b_1 & b_2 & \cdots \end{array} \qquad b_1 = \frac{a_{n-1}a_{n-2} - a_n a_{n-3}}{a_{n-1}}, \quad b_2 = \frac{a_{n-1}a_{n-4} - a_n a_{n-5}}{a_{n-1}} .

Missing entries are zeros, and a row may be scaled by a positive number. The rule (from memory, not proved here): the number of right-half-plane roots equals the number of sign changes in the first column.

Worked example 1: the cruise PI loop. For 10s3+11s2+(1+Kp)s+Ki10s^3 + 11s^2 + (1+K_p)s + K_i the first column is 10, 11, (11(1+Kp)−10Ki)/11, Ki10,\ 11,\ \bigl(11(1+K_p) - 10K_i\bigr)/11,\ K_i. No sign change needs Ki>0K_i \gt 0 and

Ki<1.1 (1+Kp),K_i \lt 1.1\,(1 + K_p) ,

which is 13.3’s edge with no root finding. With a derivative gain the 1111 becomes 11+Kd11 + K_d and the edge (11+Kd)(1+Kp)/10(11 + K_d)(1+K_p)/10: at Kp=10K_p = 10 it is 12.112.1 for Kd=0K_d = 0 and 18.718.7 for Kd=6K_d = 6 (both reported by the lab’s cruise PI and PID plants, at ω=1.0488\omega = 1.0488). Derivative damping buys room for more integral gain.

The lab’s plant for the table is L=Ki/(s(10s2+11s+5))L = K_i/\bigl(s(10s^2 + 11s + 5)\bigr) (Kp=4K_p = 4, gain K=KiK = K_i):

KiK_is1s^1 entrychangesrootsζ\zeta of pair
0.50.54.54554.545500−0.1353, −0.4823±0.3699i-0.1353,\ -0.4823 \pm 0.3699i0.79350.7935
223.18183.181800−0.7876, −0.1562±0.4791i-0.7876,\ -0.1562 \pm 0.4791i0.31000.3100
550.45450.454500−1.0696, −0.0152±0.6835i-1.0696,\ -0.0152 \pm 0.6835i0.02220.0222
5.55.5zero row−1.1, ±0.7071i-1.1,\ \pm 0.7071i00
66−0.4545-0.454522−1.1282, +0.0141±0.7291i-1.1282,\ +0.0141 \pm 0.7291i−0.0193-0.0193

At Ki=6K_i = 6 the column 10,11,−0.4545,610, 11, -0.4545, 6 has two changes: two right-half-plane roots, counted without computing them.

Special case 1: a zero pivot. Replace a zero first entry (with a nonzero rest of the row) by ε>0\varepsilon \gt 0, continue, and read signs as ε→0+\varepsilon \to 0^+. For s4+2s3+s2+2s+1s^4 + 2s^3 + s^2 + 2s + 1 the rows are [1,1,1][1,1,1], [2,2][2,2], [ε,1][\varepsilon, 1], then (2ε−2)/ε→(2\varepsilon - 2)/\varepsilon \to negative, then 11: two changes, two unstable roots. (In the lab type 0 for num and 1 2 1 2 1 for den: roots 0.2071±0.9783i0.2071 \pm 0.9783i, −1.8832-1.8832, −0.5310-0.5310.)

Special case 2: a row of zeros means roots in symmetric positions. Form the auxiliary polynomial from the row above; its roots are roots of the original. Replace the zero row by the coefficients of its derivative and continue. At Ki=5.5K_i = 5.5 the zero row is s1s^1, the row above is [11,5.5][11, 5.5], and 11s2+5.5=011s^2 + 5.5 = 0 gives ±0.7071i\pm 0.7071i, leaving −1.1-1.1. A zero row with no sign changes is the edge: marginal.

Worked example 2: a gain range. For L=K/(s(s+1)(s+2))L = K/\bigl(s(s+1)(s+2)\bigr), χ=s3+3s2+2s+K\chi = s^3 + 3s^2 + 2s + K has rows [1,2][1, 2], [3,K][3, K], [(6−K)/3][(6-K)/3], [K][K]: stable for 0<K<60 \lt K \lt 6. At K=6K = 6 the auxiliary polynomial 3s2+63s^2 + 6 gives ω=2=1.4142\omega = \sqrt2 = 1.4142; the lab agrees, and at K=1K = 1 gives −2.3247-2.3247, −0.3376±0.5623i-0.3376 \pm 0.5623i (ζ=0.5148\zeta = 0.5148).

The root locus

The root locus plots all roots of D(s)+K N(s)=0D(s) + K\,N(s) = 0 as KK runs from 00 to ∞\infty. A point ss is on it when K=−D(s)/N(s)K = -D(s)/N(s) is a positive real number: the magnitude K=∣D∣/∣N∣K = |D|/|N| gives the gain there, and the angle condition ∠N(s)−∠D(s)=180∘\angle N(s) - \angle D(s) = 180^\circ picks the points. Let DD have degree nn with poles pip_i and NN degree mm with zeros zjz_j.

Rules. The first three are from memory; the rest are derived.

  1. There are nn branches, symmetric about the real axis.
  2. They start at the poles (K=0K = 0); mm end at the zeros and n−mn - m go to infinity.
  3. A real point is on the locus if the number of real poles and zeros to its right is odd.
  4. Asymptotes. For large ∣s∣|s|, D/N≈(s−σ)n−mD/N \approx (s - \sigma)^{n-m} times a constant, with σ=(∑pi−∑zj)/(n−m)\sigma = (\sum p_i - \sum z_j)/(n-m) (compare the sn−m−1s^{n-m-1} coefficients). The locus needs (s−σ)n−m≈−K(s-\sigma)^{n-m} \approx -K, negative real, so the branches leave along lines through the centroid σ\sigma at angles (2k+1) 180∘/(n−m)(2k+1)\,180^\circ/(n-m).
  5. Breakaway and break-in. Where branches meet on the axis, K(s)=−D/NK(s) = -D/N is stationary: D N′−D′N=0D\,N' - D'N = 0. Real roots with K>0K \gt 0 are the candidates.
  6. Imaginary-axis crossings are the Routh edge: KK from the zero row, ω\omega from the auxiliary polynomial.

By Vieta, when n−m≥2n - m \ge 2 the sn−1s^{n-1} coefficient of D+KND + KN is independent of KK, so the sum of the closed-loop poles is constant. The rules give a sketch; the lab solves χ=0\chi = 0 on a grid of KK for the numbers (the leading coefficient of DD rescales KK but not the shape).

Locus 1, by hand: K/(s(s+1)(s+2))K/\bigl(s(s+1)(s+2)\bigr). Segments [−1,0][-1, 0] and (−∞,−2](-\infty, -2]; centroid −1-1, angles 60∘,180∘,300∘60^\circ, 180^\circ, 300^\circ. Breakaway: 3s2+6s+2=03s^2 + 6s + 2 = 0 gives −0.4226-0.4226 (on a segment) or −1.5774-1.5774 (not), with K=−D(s)=2/(33)=0.3849K = -D(s) = 2/(3\sqrt3) = 0.3849. Crossing at K=6K = 6, ω=2\omega = \sqrt2. The lab prints the same. Below 0.38490.3849 all roots are real; above it two become a pair that bends toward the 60∘60^\circ lines and crosses at ±1.4142i\pm1.4142i.

◆ Interactive Root locus of the cruise PI loop (Kp = 4, gain K = Ki): drag K through the edge at 5.5 and watch the pair cross the imaginary axis while the Routh array loses its last positive entry

Loading simulation…

Locus 2: the cruise plant with engine lag, K=KpK = K_p. L=K/((10s+1)(s+1))L = K/\bigl((10s+1)(s+1)\bigr) has poles −0.1-0.1, −1-1 and n−m=2n - m = 2: segment [−1,−0.1][-1, -0.1], centroid −0.55-0.55, asymptotes vertical. The pole sum is fixed at −1.1-1.1, so after leaving the axis the pair stays on Re s=−0.55\text{Re}\,s = -0.55, which is why 13.2 found −0.55±ωi-0.55 \pm \omega i at every gain. Breakaway: s=−0.55s = -0.55 at K=2.025K = 2.025, matching 13.2’s discriminant (the lab prints −0.5500-0.5500, 2.02502.0250). At K=4K = 4 the roots are −0.55±0.4444i-0.55 \pm 0.4444i (ζ=0.7778\zeta = 0.7778, ωn=0.7071\omega_n = 0.7071); at K=19K = 19, −0.55±1.3029i-0.55 \pm 1.3029i (ζ=0.3889\zeta = 0.3889, ωn=1.4142\omega_n = 1.4142). There is no crossing: the lab reports it stable over its whole range.

Locus 3: PI with K=KiK = K_i at Kp=4K_p = 4. Poles 00 and −0.55±0.4444i-0.55 \pm 0.4444i, no zeros: segment (−∞,0](-\infty, 0], centroid (0−1.1)/3=−0.3667(0 - 1.1)/3 = -0.3667, angles ±60∘\pm60^\circ, 180∘180^\circ. Rule 5: 30s2+22s+530s^2 + 22s + 5 has discriminant 484−600<0484 - 600 \lt 0, so no breakaway. The complex branches run out along ±60∘\pm60^\circ, into the right half plane, and must cross: at Ki=5.5K_i = 5.5, ω=0.7071\omega = 0.7071. The table above is this locus sampled. Everything the integral gain does to the pair is paid out of damping.

The yaw damper of Module 12

In 12.6 Lateral-Directional Modes the damper δr=−Kr\delta_r = -Kr changes one column of the 4×44\times4 matrix: Acl=A−K bre3TA_{\text{cl}} = A - K\,\mathbf b_r\mathbf e_3^{\mathsf T} with br=(−4.2952, 0.3568, 5.0643, 0)\mathbf b_r = (-4.2952,\ 0.3568,\ 5.0643,\ 0) from 12.6’s lab cell. A rank-one change makes the characteristic polynomial linear in KK, χK=p0+K q\chi_K = p_0 + K\,q. Numerically, rounded to five places (the lab uses these),

p0=s4+10.35208s3+14.61023s2+51.46477s+0.34816,q=5.06431s3+47.97641s2+6.54292s+16.07456.p_0 = s^4 + 10.35208s^3 + 14.61023s^2 + 51.46477s + 0.34816, \quad q = 5.06431s^3 + 47.97641s^2 + 6.54292s + 16.07456 .

The open-loop poles are roll −9.3790-9.3790, Dutch roll −0.4832±2.2898i-0.4832 \pm 2.2898i, spiral −0.0068-0.0068; the zeros of qq are −9.3717-9.3717 (next to the roll pole, so roll barely reacts) and a lightly damped pair −0.0509±0.5797i-0.0509 \pm 0.5797i. With n−m=1n - m = 1 there is one asymptote at 180∘180^\circ. At K=0.5K = 0.5 the lab gives −9.3818-9.3818, −0.1731-0.1731, −1.6647±1.5465i-1.6647 \pm 1.5465i (ζ=0.7326\zeta = 0.7326), 12.6’s table. The locus explains it:

  • Breakaway at K=0.7165K = 0.7165, s=−2.1634s = -2.1634: the Dutch-roll pair splits (12.6 said “about 0.720.72”). The pair’s best damping is just before, ζ=0.9776\zeta = 0.9776 at K=0.7K = 0.7.
  • Break-in at K=0.9665K = 0.9665, s=−0.6030s = -0.6030: the branch moving right meets the spiral root moving left, and at K=1K = 1 they are a pair −0.5781±0.1575i-0.5781 \pm 0.1575i.
  • That pair is bound for the zeros. At K=2K = 2 it is −0.2943±0.4932i-0.2943 \pm 0.4932i (ζ=0.5124\zeta = 0.5124), at K=3K = 3 −0.2129±0.5329i-0.2129 \pm 0.5329i (ζ=0.3711\zeta = 0.3711), ending where ζ=0.0509/0.5820=0.0875\zeta = 0.0509/0.5820 = 0.0875. Past K≈0.7K \approx 0.7, more gain trades the fast Dutch roll for a slow, poorly damped oscillation near 0.580.58 rad/s.

The loop stays stable over the slider range. Reading the damping off the locus, ζ≈0.7\zeta \approx 0.7 is near K=0.5K = 0.5.

◆ Interactive The yaw damper locus: K in seconds from 0 to 3, breakaway at 0.7165 and break-in at 0.9665

Loading simulation…

Delay: a transfer function that is not a polynomial

A delay multiplies LL by e−sθe^{-s\theta}, which is not rational, so χ\chi has infinitely many roots and the rules above do not apply. 13.1 used the phase condition: for 1/(10s+1)1/(10s+1) with θ=1\theta = 1 s the loop is critical when arctan⁡(10ω)+ω=π\arctan(10\omega) + \omega = \pi, at ω=1.632\omega = 1.632 and Kc=1+100ω2=16.351K_c = \sqrt{1 + 100\omega^2} = 16.351 (recomputed: ω=1.6320\omega = 1.6320, Kc=16.3506K_c = 16.3506).

To use polynomial tools, approximate. The first-order Pade form (from memory) is e−s≈(1−s/2)/(1+s/2)e^{-s} \approx (1 - s/2)/(1 + s/2), which matches the series through s2s^2 (1−s+s2/2−s3/41 - s + s^2/2 - s^3/4 against −s3/6-s^3/6). Then χ=5s2+(10.5−0.5K)s+1+K\chi = 5s^2 + (10.5 - 0.5K)s + 1 + K, stable for K<21K \lt 21, with ω2=(1+K)/5\omega^2 = (1+K)/5, ω=2.0976\omega = 2.0976 at the edge: the lab gives 21.000021.0000 at 2.09762.0976 rad/s, 28%28\% above the exact 16.35116.351, an unsafe error. The second-order form (12−6s+s2)/(12+6s+s2)(12 - 6s + s^2)/(12 + 6s + s^2) (from memory) gives 16.486616.4866 at 1.64561.6456 rad/s, 0.8%0.8\% above.

Two lessons. The Pade form puts a zero at +2+2, in the right half plane (a non-minimum-phase zero): the signature of a delay, since the locus must end on it and the gain is capped. And the right structure can still miss the number that matters: at K=14.5K = 14.5 the first form gives −0.3250±1.7304i-0.3250 \pm 1.7304i (ζ=0.1846\zeta = 0.1846), the second −0.0933±1.5866i-0.0933 \pm 1.5866i (ζ=0.0587\zeta = 0.0587) with −7.3634-7.3634; 13.1 saw ringing that decays at 14.514.5 and barely decays at 1616.

More worked examples

Counting unstable roots. For s3+s2+2s+8s^3 + s^2 + 2s + 8 the rows are [1,2][1, 2], [1,8][1, 8], [−6][-6], [8][8]: the column 1,1,−6,81, 1, -6, 8 has two changes. Check: (s+2)(s2−s+4)(s+2)(s^2 - s + 4), roots −2-2, 0.5±1.9365i0.5 \pm 1.9365i.

Concept checks

✅ Concept check

The loop with $\chi(s) = s^3 + 3s^2 + 2s + K$ (the plant $K/(s(s+1)(s+2))$). What is the largest $K$ for which every closed-loop pole is in the open left half plane?

✅ Concept check

Applying the Routh array to $s^3 + s^2 + 2s + 8$ gives the first column $1, 1, -6, 8$. What does that say?

✅ Concept check

An open-loop transfer function has poles at $0$, $-2$ and $-4$ and no zeros. Where is the centroid of the asymptotes?

✅ Concept checkselect all that apply

Select every statement that is true.

Playground and exercise

The first cell finds loci numerically with a root finder and bisects for the edge gain, the second asks you to write the Routh test itself.

🐍 PythonPlayground: roots over a gain grid, the edge by bisection, and the cruise loop at K = 4 and 19
⌘/Ctrl + Enter
🧪 Coding exerciseExercise: the Routh first column, the unstable-root count, and the centroid
⌘/Ctrl + Enter

Problems

✏️ Pen & paperThe PID edge with a derivative term●○○

The cruise plant with the engine lag is under PID control with Kp=4K_p = 4 and Kd=9K_d = 9. Find the largest stable KiK_i and the frequency of oscillation there.

Show worked solution

The polynomial is 10s3+20s2+5s+Ki10s^3 + 20s^2 + 5s + K_i. The array is [10,5][10, 5], [20,Ki][20, K_i], [(100−10Ki)/20][(100 - 10K_i)/20], [Ki][K_i], so 0<Ki<100 \lt K_i \lt 10. At Ki=10K_i = 10 the auxiliary polynomial is 20s2+10=020s^2 + 10 = 0, ω=0.5=0.7071\omega = \sqrt{0.5} = 0.7071 rad/s. Check: 10s3+20s2+5s+10=(s2+0.5)(10s+20)10s^3 + 20s^2 + 5s + 10 = (s^2 + 0.5)(10s + 20), so the third root is −2-2. In general the edge is (11+Kd)(1+Kp)/10=20⋅5/10=10(11 + K_d)(1+K_p)/10 = 20 \cdot 5/10 = 10 and ω2=(1+Kp)/10\omega^2 = (1+K_p)/10 does not depend on KdK_d.

✏️ Pen & paperA locus with a zero●●○

Sketch the locus of L=K(s+3)/(s(s+1)(s+2))L = K(s + 3)/\bigl(s(s+1)(s+2)\bigr): real-axis segments, asymptotes and centroid, and the breakaway point (as a root of a cubic). Then use Routh to show the loop is stable for every K>0K \gt 0.

Show worked solution

Poles 0,−1,−20, -1, -2, zero −3-3: n=3n = 3, m=1m = 1. Real-axis points with an odd count of poles and zeros to their right: [−1,0][-1, 0] and [−3,−2][-3, -2]. Two branches go to infinity: angles 90∘90^\circ and 270∘270^\circ, centroid (0−1−2+3)/2=0(0 - 1 - 2 + 3)/2 = 0, so the asymptotes run along the imaginary axis. The third branch ends on the zero at −3-3. Breakaway: DN′−D′N=(s3+3s2+2s)−(3s2+6s+2)(s+3)=−(2s3+12s2+18s+6)D N' - D'N = (s^3 + 3s^2 + 2s) - (3s^2 + 6s + 2)(s+3) = -(2s^3 + 12s^2 + 18s + 6), so s3+6s2+9s+3=0s^3 + 6s^2 + 9s + 3 = 0, whose roots are −0.4679-0.4679, −1.6527-1.6527 and −3.8794-3.8794. Only −0.4679-0.4679 is on the locus (on [−1,0][-1, 0]), at K=−D/N=0.1506K = -D/N = 0.1506. For the other roots K<0K \lt 0. Stability: χ=s3+3s2+(2+K)s+3K\chi = s^3 + 3s^2 + (2 + K)s + 3K has array [1,2+K][1, 2 + K], [3,3K][3, 3K], [(3(2+K)−3K)/3=2][(3(2+K) - 3K)/3 = 2], [3K][3K]. The first column 1,3,2,3K1, 3, 2, 3K is positive for all K>0K \gt 0. The pair approaches the imaginary axis from the left as KK grows (centroid 00) without ever crossing it, and the sum of the poles stays −3-3, so the third root tends to −3-3. A numerical check: at K=10K = 10 the roots are −0.1560±3.3372i-0.1560 \pm 3.3372i and −2.6879-2.6879.

✏️ Pen & paperA zero row●●○

Apply the Routh test to s4+3s3+3s2+3s+2s^4 + 3s^3 + 3s^2 + 3s + 2. What happens in row s1s^1, and what do you conclude about the roots?

Show worked solution

Rows: [1,3,2][1, 3, 2], [3,3][3, 3], [2,2][2, 2], and then (2⋅3−3⋅2)/2=0(2\cdot 3 - 3 \cdot 2)/2 = 0 for the s1s^1 row, which is entirely zero. The auxiliary polynomial from the s2s^2 row is 2s2+2=02s^2 + 2 = 0, so s=±is = \pm i are roots. Replacing the zero row by the derivative 4s4s gives the column 1,3,2,4,21, 3, 2, 4, 2 with no sign change, so the other two roots are in the left half plane. Indeed (s2+1)(s2+3s+2)(s^2 + 1)(s^2 + 3s + 2) has roots ±i,−1,−2\pm i, -1, -2: marginally stable, with an undamped oscillation at ω=1\omega = 1.

✏️ Pen & paperPadé versus exact delay●●●

The plant 1/(2s+1)1/(2s + 1) with a delay θ=0.5\theta = 0.5 s is under proportional gain KK. Use the first-order Pade form e−0.5s≈(1−s/4)/(1+s/4)e^{-0.5s} \approx (1 - s/4)/(1 + s/4) to find the critical gain and its frequency. The exact phase condition is arctan⁡(2ω)+0.5ω=π\arctan(2\omega) + 0.5\omega = \pi. Which is larger?

Show worked solution

χ=(2s+1)(1+s/4)+K(1−s/4)=s2/2+(2.25−K/4)s+1+K\chi = (2s+1)(1 + s/4) + K(1 - s/4) = s^2/2 + (2.25 - K/4)s + 1 + K. The ss coefficient vanishes at K=9K = 9, where ω2=(1+K)/0.5=20\omega^2 = (1 + K)/0.5 = 20, so ω=4.4721\omega = 4.4721 rad/s. The exact condition arctan⁡(2ω)+0.5ω=π\arctan(2\omega) + 0.5\omega = \pi is solved numerically by ω=3.431\omega = 3.431, giving Kc=1+4ω2=6.935K_c = \sqrt{1 + 4\omega^2} = 6.935. So the Pade value 99 is the larger, about 30%30\% too generous: it would call stable a loop that is not, the same direction of error as the 2121 against 16.35116.351 of the lesson.

What comes next

Everything here needed a polynomial, a model. The delay was the gap: Pade moved the critical gain from 2121 to 16.486616.4866 against the exact 16.35116.351, while the exact answer came from a phase condition on the frequency response L(iω)L(i\omega) that needed no polynomial. Lesson 13.5 (13.5 Frequency Response & Bode Plots) takes that route: gain and phase against frequency as a Bode plot, with the margin to instability read off it, for a model or a measured system.

Key Takeaways

  • A loop is stable when every root of χ(s)=D+KN\chi(s) = D + KN is in the open left half plane; the imaginary axis is marginal, not safe.
  • Routh counts right-half-plane roots as first-column sign changes (from memory, unproved). A zero pivot takes ε\varepsilon; a zero row uses the auxiliary polynomial, and at an edge its roots are the oscillation.
  • The cruise PI loop is stable for 0<Ki<(11+Kd)(1+Kp)/100 \lt K_i \lt (11 + K_d)(1 + K_p)/10: 5.55.5 at Kp=4K_p = 4, 12.112.1 at Kp=10K_p = 10, 18.718.7 with Kd=6K_d = 6.
  • Root locus rules: nn branches, odd-count real segments, asymptotes at (2k+1)180∘/(n−m)(2k+1)180^\circ/(n-m) through σ=(∑p−∑z)/(n−m)\sigma = (\sum p - \sum z)/(n-m), breakaway where DN′−D′N=0DN' - D'N = 0, crossings from Routh. For n−m≥2n - m \ge 2 the pole sum is constant.
  • Cruise plant with lag: breakaway s=−0.55s = -0.55 at K=2.025K = 2.025, vertical locus. PI at Kp=4K_p = 4: crossing at Ki=5.5K_i = 5.5, ω=0.7071\omega = 0.7071.
  • Yaw damper: breakaway K=0.7165K = 0.7165 (ζ\zeta peak 0.97760.9776 at K=0.7K = 0.7), break-in 0.96650.9665, then a slower pair heads for the zeros at −0.0509±0.5797i-0.0509 \pm 0.5797i.
  • Pade (1,1) puts the delay’s critical gain at 2121, Pade (2,2) at 16.486616.4866, the exact value is 16.35116.351.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.

Where this leads

These later lessons build directly on this one: