Skip to content
Tutorials Center
My learning

Topics

Lesson 37 of 37 · Projects and Practice

Control Systems Practice Problems

Ten worked control-systems problems covering second-order response, steady-state error, Routh stability, root locus, gain margin, PID tuning, state space, LQR and discretization.

The best way to find out what you really understand is to solve problems without looking at the lesson. This page gives ten problems, one for each of the main skills from the second half of the course. Try each on paper first; the code blocks compute the answers so you can check yourself, and the explanations show how to get them by hand.

The problems

  1. Second-order response. A unity-feedback loop has G(s) = 10/(s(s+2)). Find the closed-loop poles, ωn, ζ and the percent overshoot of the step response.
  2. Steady-state error. For the same loop, find the velocity error constant Kv and the steady-state error for a unit ramp input.
  3. Routh stability. For the characteristic equation s³ + 6s² + 5s + K = 0, find the range of K for stability.
  4. Root locus. For L(s) = K/(s(s+4)(s+6)), find the centroid and asymptote angles, and the gain at which the locus crosses the imaginary axis.
  5. Gain margin. For L(s) = 10/(s(s+1)(s+5)), find the phase crossover frequency and the gain margin.
  6. PID tuning. A plant under proportional control oscillates steadily at Ku = 12 with period Tu = 2.5 s. Give the Ziegler-Nichols PID parameters Kp, Ti, Td, and Ki, Kd.
  7. State space. The system x' = Ax + Bu has A = [[0,1,0],[0,0,1],[-6,-11,-6]], B = [0,0,1]ᵀ. Is it controllable? What are its poles?
  8. Pole placement. You need a 2% settling time of 1 s and ζ = 0.7. Where should the dominant closed-loop poles be?
  9. Discretization. A continuous pole at s = -3 is sampled with T = 0.1 s. Find the discrete pole exactly, with forward Euler and with Tustin.
  10. LQR. The scalar system x' = x + 2u has cost ∫ (3x² + u²) dt. Find the optimal feedback gain and the closed-loop pole.

Solutions: part 1 (problems 1 to 5)

Python Check your answers to problems 1 to 5

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

How to get them by hand:

  1. The closed loop is 10/(s² + 2s + 10), so ωn = √10 = 3.16, 2ζωn = 2 gives ζ = 0.316, poles -1 ± 3j, and overshoot exp(-πζ/√(1-ζ²)) ≈ 35%.
  2. Kv = lim s·G(s) = 10/2 = 5, so the unit-ramp error is 1/Kv = 0.2.
  3. Routh's first column is 1, 6, (30 - K)/6, K. All positive needs 0 < K < 30; at K = 30 there is an imaginary pair ±j√5.
  4. n = 3, m = 0: centroid (0 - 4 - 6)/3 = -3.33, angles 60°, 180°, 300°. Crossing: s³ + 10s² + 24s + K gives ω² = 24, K = 10·24 = 240.
  5. Phase crossover where -90° - arctan ω - arctan(ω/5) = -180°, which happens at ω = √5 = 2.24. There |L| = 10/(ω·√(1+ω²)·√(25+ω²)) ≈ 0.333, so the gain margin is 3 (9.5 dB); equivalently K_crit = 30.

Solutions: part 2 (problems 6 to 10)

Python Check your answers to problems 6 to 10

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

How to get them by hand:

  1. Kp = 0.6·Ku = 7.2, Ti = Tu/2 = 1.25 s, Td = Tu/8 = 0.3125 s, so Ki = Kp/Ti = 5.76, Kd = Kp·Td = 2.25.
  2. The matrix is in controllable canonical form, so it is controllable (rank 3). The characteristic polynomial is s³ + 6s² + 11s + 6 = (s+1)(s+2)(s+3), so the poles are -1, -2, -3.
  3. σ = 4/ts = 4, and ζ = 0.7 gives ωd = σ·√(1-ζ²)/ζ = 4.08. The poles are -4 ± 4.08j (so ωn = 5.71).
  4. Exact z = e^(-0.3) = 0.7408; forward Euler z = 1 + sT = 0.7; Tustin z = (1 - 0.15)/(1 + 0.15) = 0.7391. Tustin is much closer here, as the sampling is fast relative to the pole.
  5. The Riccati equation 2aP - (b²/r)P² + q = 0 with a = 1, b = 2, q = 3, r = 1 is 2P - 4P² + 3 = 0, whose positive root is P = (1 + √13)/4 = 1.1514. The gain is k = bP/r = 2.303 and the closed-loop pole is a - bk = -3.606.

Where to go from here

If any answer surprised you, go back to the lesson for that skill:

Problem Review
1, 2 Time-domain analysis (second-order systems, error constants)
3 Stability and the Routh-Hurwitz criterion
4 Root locus rules
5 Nyquist and margins
6 Ziegler-Nichols tuning
7 Controllability
8 Pole placement
9 Discretization
10 LQR

Bigger challenges

  1. Add a lead compensator to problem 1's loop to reach 50° phase margin, then discretize it with Tustin at T = 0.02 s and check the digital loop in simulation.
  2. Take the cruise-control project and implement the controller in velocity form with a rate limit.
  3. Make a three-state observer for the problem 7 system when only the first state is measured, and find the fastest observer poles that keep the estimate noise below a threshold of your choice.
  4. Re-do the pendulum project with a position-and-angle reference tracker.

Check your understanding

Quick check · 3 questions

of 3 answered You scored / 3 · perfect!

  1. 1. A loop has a gain margin of 3 and is currently at K = 10. At what gain does it become unstable?

    The gain margin is the factor by which K can increase before instability, so K_crit = 3 * 10 = 30.
  2. 2. In the scalar LQR problem, increasing r (the control penalty) will:

    A larger r makes control effort expensive, so the optimal feedback is weaker.
  3. 3. A zero steady-state error for a ramp input requires the loop to be at least which type?

    A type-1 loop has a constant ramp error 1/Kv; a type-2 loop (two integrators) tracks a ramp with zero error.

Summary

  • Ten problems cover time response, error constants, Routh, root locus, margins, PID, state space, pole placement, discretization and LQR.
  • Always verify analytic answers numerically: NumPy, SciPy and a few lines of simulation are your safety net.
  • Revisit the lesson linked from any problem you missed, then try the bigger challenges.
  • You now have the full classical and modern toolbox: model, analyze, design, discretize and test.

Next: Control Systems tutorial index