Skip to content
Tutorials Center
My learning

Topics

Lesson 29 of 36 · Sampling and Z-transform

The Z-transform and Its Region of Convergence

Learn the z-transform of discrete-time signals, the region of convergence, key pairs and properties, and how the unit circle connects it to the DTFT, checked in Python.

The z-transform is to discrete-time signals what the Laplace transform is to continuous ones. It converts difference equations into algebra, lets you describe a digital system with a transfer function H(z), and shows stability through the location of poles. If you have been through the Laplace transform, you will recognize almost every idea.

Definition

The bilateral z-transform of a sequence x[n] is

X(z) = Σ from n = -∞ to ∞ of x[n] z^(-n), where z is a complex variable.

The unilateral version sums from n = 0. Writing z = r e^(jω) turns the sum into the DTFT of x[n] r^(-n). When the unit circle (r = 1) lies inside the region of convergence, X(e^(jω)) is the DTFT of x[n]. That is the discrete counterpart of "the imaginary axis gives the Fourier transform" for Laplace.

A first example: the geometric sequence

For x[n] = aⁿ u[n]:

X(z) = Σ from n = 0 to ∞ of (a/z)ⁿ = 1 / (1 - a z^(-1)) = z / (z - a), for |z| > |a|.

The pole is at z = a and the ROC is outside the circle of radius |a|, the usual shape for right-sided sequences. The sequence is stable (absolutely summable) when |a| < 1, in which case the ROC contains the unit circle.

Python Geometric series versus the closed form

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.

ROC rules

  • The ROC is a ring (annulus) centered on the origin: r1 < |z| < r2.
  • It contains no poles.
  • A right-sided sequence has an ROC outside its outermost pole (extending to infinity if causal).
  • A left-sided sequence has an ROC inside its innermost pole.
  • A finite-length sequence converges everywhere except possibly z = 0 and z = ∞.
  • A causal LTI system is stable exactly when all poles are inside the unit circle.

Common pairs

x[n] X(z) ROC
δ[n] 1 all z
u[n] z/(z - 1) |z| > 1
aⁿ u[n] z/(z - a) |z| > |a|
n aⁿ u[n] a z / (z - a)² |z| > |a|
-aⁿ u[-n-1] z/(z - a) |z| < |a|
cos(ω0 n) u[n] z(z - cos ω0) / (z² - 2z cos ω0 + 1) |z| > 1
rⁿ cos(ω0 n) u[n] z(z - r cos ω0) / (z² - 2 r z cos ω0 + r²) |z| > r

Notice the pairs for aⁿ u[n] and -aⁿ u[-n-1]: identical formula, different ROC, exactly as with Laplace.

Properties

Property Time domain z-domain
Linearity a x + b y a X + b Y
Time shift x[n - k] z^(-k) X(z)
Scaling by aⁿ aⁿ x[n] X(z/a)
Time reversal x[-n] X(1/z)
Differentiation in z n x[n] -z dX/dz
Convolution x * y X(z) Y(z)

The time-shift rule is what makes difference equations easy: a one-sample delay is multiplication by z^(-1), which block diagrams draw as a box labeled z^(-1).

From X(z) to the frequency response

Evaluate H(z) on the unit circle to get the frequency response. For the finite sequence h = [1, 2, 1] (a smoothing filter), H(z) = 1 + 2z^(-1) + z^(-2) = (1 + z^(-1))². It has a double zero at z = -1, so the response is zero at ω = π: this filter kills the fastest-changing component.

Python Unit circle evaluation equals the DTFT

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.

The relation to the Laplace transform

Sampling a continuous-time system with period Ts maps an s-plane pole at s = p to a z-plane pole at z = e^(p Ts). The left half of the s-plane maps to the interior of the unit circle, and the imaginary axis maps to the unit circle itself. This is why "left half-plane" and "inside the unit circle" are the same stability criterion in two coordinates.

In control and DSP software, polynomials are often written in z^(-1) (as scipy.signal.lfilter does), while pole-zero plots use z. Converting between them multiplies numerator and denominator by a power of z; the pole-zero locations from np.roots of the z^(-1) polynomial reversed give you z-plane locations. Always state which form you use.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. What is the z-transform of x[n] = (0.5)ⁿ u[n]?

    Summing the geometric series gives 1 / (1 - 0.5 z^(-1)) = z / (z - 0.5), valid outside the circle |z| = 0.5.
  2. 2. A causal system has poles at z = 0.9 and z = -0.5. Is it stable?

    Stability for a causal system needs all poles inside the unit circle: |0.9| < 1 and |-0.5| < 1.
  3. 3. What does the factor z^(-1) represent in a block diagram?

    By the time-shift property, x[n - 1] has z-transform z^(-1) X(z).
  4. 4. How do you obtain the frequency response from H(z)?

    H(e^(jω)) is the DTFT of the impulse response, provided the unit circle is in the ROC.

Summary

  • X(z) = Σ x[n] z^(-n); the ROC is an annulus and is part of the answer.
  • Right-sided sequences converge outside their outermost pole, left-sided inside the innermost.
  • A causal LTI system is stable when all poles are inside the unit circle.
  • A delay of one sample is z^(-1); convolution becomes a product.
  • The DTFT is X(z) on the unit circle.

Next: Inverse Z-transform and Difference Equations