Quiz · Hard
Applied problems on root locus gain limits, phase margin, lead compensation, Nyquist counting and controllability.
For the system above, the peak time is about
0.314 s
0.628 s
0.363 s
1 s
omega_d = 10 sqrt(1 - 0.25) = 8.66 rad/s; tp = pi / 8.66 = 0.363 s. 0.314 s would use omega_n instead of omega_d.
A unity-feedback loop has . The maximum K for stability is
8
24
48
6
Characteristic polynomial s^3 + 6s^2 + 8s + K. Routh s^1 row = (6*8 - K)/6, so K < 48. At K = 48 the locus crosses at omega = sqrt(8) = 2.83.
For the same system, the asymptote centroid is
-6
-3
0
-2
Centroid = (sum of poles - sum of zeros)/(n - m) = (0 - 2 - 4)/3 = -2.
A loop transfer function has a gain crossover frequency where its phase is -120 degrees. The phase margin is
60 degrees
120 degrees
-60 degrees
30 degrees
PM = 180 + (-120) = 60 degrees. The approximate damping ratio is about 0.6.
A phase margin of 45 degrees corresponds roughly to a damping ratio of
0.7
0.45
0.2
0.9
Use zeta ~ PM / 100 for PM up to about 60 degrees.
Which closed-loop pole location gives the fastest decay with no oscillation?
-1 + j10
+1
-10 (real axis)
0 + j10
A real pole far into the left half-plane decays fast and does not oscillate. -1 + j10 rings; +1 is unstable; +/- j10 oscillates forever.
A lead compensator with zero at 2 rad/s and pole at 20 rad/s provides a maximum phase lead of about
90 degrees
18 degrees
45 degrees
54.9 degrees
alpha = 10, sin(phi) = 9/11 = 0.818, phi = 54.9 degrees, at sqrt(2 * 20) = 6.3 rad/s.
The Nyquist plot of a stable open-loop system encircles -1 once clockwise. The closed loop has
1 RHP pole
no RHP poles
2 RHP poles
an undefined number of poles
Z = N + P = 1 + 0 = 1 closed-loop right-half-plane pole, so the system is unstable.
. The system poles are
-1 and -6
-2 and -3
2 and 3
-5 and -6
det(sI - A) = s^2 + 5s + 6 = (s + 2)(s + 3). Both roots are negative, so it is stable.
With and , the controllability matrix is
[[0, 0], [1, 0]], rank 1, so not controllable
[[1, 0], [0, 1]], rank 2, so observable
[[0, 1], [1, 0]], full rank, so controllable
[[0, 1], [0, 0]], rank 1, so not controllable
AB = [1, 0]^T, so [B AB] = [[0, 1], [1, 0]] with determinant -1: rank 2. (This is a double integrator.)
A controller's integrator keeps accumulating while the actuator is saturated. The resulting large overshoot is called
derivative kick
gain compression
limit cycling
integral windup
Anti-windup schemes clamp the integrator or back-calculate it when saturation occurs.
A plant has a right-half-plane zero. In its step response you expect
an initial response in the wrong direction (undershoot)
a faster response with no overshoot
a steady-state error of zero
an unstable open loop
RHP zeros give non-minimum-phase behavior: the output first moves opposite to its final value.
A system with transfer function has and equal to
100 rad/s and 0.1
10 rad/s and 0.5
10 rad/s and 1
10 rad/s and 0.05
omega_n^2 = 100 so omega_n = 10; 2 zeta omega_n = 10 so zeta = 0.5.
A unity-feedback type 0 plant has Kp = 9. The step-input steady-state error is
9%
0%
90%
10%
e_ss = 1 / (1 + Kp) = 1/10 = 10%. Zero error requires an integrator in the loop.