Skip to content
Tutorials Center
My learning

Topics

Lesson 36 of 36 Ā· Projects and Practice

Practice Problems for Signals and Systems

Eight worked practice problems covering convolution, energy, Fourier and Laplace transforms, aliasing, z-transform stability and filters, each checked with Python.

Reading about transforms is not the same as using them. This final lesson gives eight problems that mix ideas from the whole chapter range: Fourier, Laplace, sampling and the z-transform. Try each one on paper first. Then run the checking code, which verifies the answers numerically, so you can see where you went wrong.

The problems

  1. Convolution. Compute the convolution of x[n] = [1, 2, 1] and h[n] = [1, -1, 3].
  2. Energy. Find the energy of x(t) = e^(-3t) u(t), in the time domain.
  3. Fourier transform. Find X(f) for x(t) = e^(-|t|) and evaluate X(0).
  4. Inverse Laplace. Invert X(s) = (2s + 5) / ((s + 1)(s + 4)) for a causal signal.
  5. Aliasing. A 7 kHz tone is sampled at 10 kHz. At what frequency does it appear?
  6. Z-transform and stability. Is the causal system with H(z) = 1 / (1 - 0.9 z^(-1) + 0.2 z^(-2)) stable?
  7. Frequency response. For the three-point moving average y[n] = (x[n] + x[n-1] + x[n-2]) / 3, find the gain at ω = Ļ€/3 and at ω = Ļ€.
  8. Resolution. You record at 8 kHz and want to separate two tones 20 Hz apart with a rectangular window. About how many samples do you need at minimum?

Hints: problem 3 splits into two one-sided exponentials; problem 4 uses the cover-up method; problem 6 factors z² - 0.9z + 0.2; problem 7 uses H(e^(jω)) = (1 + 2 cos ω) e^(-jω) / 3.

Check problems 1 to 4

Python Verify problems 1-4

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

Program input Ā· 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.

Solutions 1 to 4

  1. Slide and add: the output has length 3 + 3 - 1 = 5. The values are [1, 1, 2, 5, 3].
  2. E = ∫ e^(-6t) dt from 0 to āˆž = 1/6. By Parseval, the same value comes from integrating |X(f)|² = 1/(9 + (2Ļ€f)²).
  3. Split at zero: e^(-t) u(t) gives 1/(1 + j2Ļ€f) and the mirrored part e^(t) u(-t) gives 1/(1 - j2Ļ€f). Their sum is 2/(1 + (2Ļ€f)²), so X(0) = 2. Because x(t) is real and even, X(f) is real and even too.
  4. A = (2(-1) + 5) / (-1 + 4) = 1 and B = (2(-4) + 5) / (-4 + 1) = 1. Therefore x(t) = (e^(-t) + e^(-4t)) u(t).

Check problems 5 to 8

Python Verify problems 5-8

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

Program input Ā· 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.

Solutions 5 to 8

  1. The alias is |7000 - 10000 Ā· round(0.7)| = |7000 - 10000| = 3 kHz. The 7 kHz tone is above the Nyquist frequency of 5 kHz, so it folds down to 3 kHz.
  2. z² - 0.9z + 0.2 = (z - 0.5)(z - 0.4). Both poles are inside the unit circle, so a causal implementation is stable.
  3. H(e^(jω)) = (1 + 2 cos ω)/3 times a phase factor. At ω = Ļ€/3, cos ω = ½, so the gain is 2/3. At ω = Ļ€, 1 + 2(-1) = -1, so the magnitude is 1/3. Note that the gain does not reach zero at ω = Ļ€; a two-point average (1 + z^(-1))/2 does.
  4. Resolution is about fs/N, so N ā‰ˆ fs/Ī”f = 8000/20 = 400 samples, a 50 ms record. In practice use a window and more samples.

Tips for self-study

  • Solve problems with two methods (a formula and a numerical check). Disagreement is a signal to look for a missing 2Ļ€ or a wrong sign.
  • Draw the pole-zero plot before you write a formula for the response.
  • Keep a one-page table of transform pairs and properties, and build it yourself from the lessons.
  • Rework a project: change a parameter and predict what happens to the plot before you run it.

Want more? Change the numbers in each TryIt and predict the new answers. For example, try problem 6 with 0.9 replaced by 1.4 and decide whether the system becomes unstable before running the code.

Check your understanding

Quick check Ā· 4 questions

of 4 answered You scored / 4 Ā· perfect!

  1. 1. What is the length of the convolution of a sequence of length 6 with one of length 4?

    The output of a full linear convolution has length L1 + L2 - 1 = 6 + 4 - 1 = 9.
  2. 2. A causal system has poles at z = 0.5 and z = -1.2. Is it stable?

    Stability of a causal system needs every pole strictly inside the unit circle.
  3. 3. A 12 kHz tone is sampled at 16 kHz. At what frequency does it appear?

    round(12/16) = 1, so the alias is |12 - 16| = 4 kHz. The Nyquist frequency is only 8 kHz.
  4. 4. Which transform would you choose to solve a linear ODE with nonzero initial conditions?

    The unilateral Laplace transform includes initial conditions in its derivative rule.

Summary

  • Convolution, energy, transform pairs, partial fractions, aliasing, pole locations and frequency responses are the everyday skills of the subject.
  • Numerical checks catch factor-of-2Ļ€ and sign errors.
  • Solve each problem twice, by formula and by simulation, until the two always agree.

Next: Back to the Signals and Systems track page