DIGITAL SIGNAL PROCESSING

Week 4: The Z-Transform

Sessions 7–8 · Oppenheim & Schafer, Ch. 3.1–3.4 · CLO1
 Week 4 Agenda
Two 75–90 minute sessions

SESSION 1 SESSION 2

Z-Transform Properties & Inverse Z-

The Z-Transform: Definition & ROC
Transform

• Motivation: generalizing the DTFT

• Core properties (linearity, shifting, convolution)
• Z-transform definition
• Poles, zeros, and pole-zero plots
• Relation to the DTFT (unit circle)
• Common Z-transform pairs
• The Region of Convergence (ROC)
• Inverse Z-transform: three methods
• ROC for right-sided, left-sided, two-sided sequences
• Partial fraction expansion, worked example
• Worked example

CLO1: Explain the fundamental theory of discrete-time signals and systems.

Textbook: Oppenheim & Schafer, Ch. 3.1–3.4 DSP · Week 4
 Why Go Beyond the DTFT?
The Z-transform generalizes the DTFT to a larger, more flexible class of signals

• The DTFT only converges for absolutely (or square) summable sequences — many useful sequences, like a growing exponential,
have no DTFT at all.

• The Z-transform introduces an extra degree of freedom — a complex variable z with adjustable magnitude — that can make the
sum converge even when the DTFT cannot.

• It also turns difference equations into algebraic equations, making system analysis (stability, causality, transfer functions) far
simpler.

• The DTFT is simply the Z-transform evaluated on the unit circle, |z| = 1 — a special case, not a separate theory.

Session 1 · Ch. 3.1 DSP · Week 4

 The Z-Transform: Definition
A power series in the complex variable z

X(z) = Σₙ₌−∞∞ x[n]·z⁻ⁿ

• z is a complex variable: z = r·e^{jω}, with magnitude r and angle ω.

• X(z) is, in general, a function defined only where this infinite sum converges — not for every z.

• Notation: x[n] ↔ X(z) (a Z-transform pair)

Session 1 · Ch. 3.1 DSP · Week 4

 Relation to the DTFT
The DTFT is the Z-transform evaluated on the unit circle

X(e^{jω}) = X(z) |_{z = e^{jω}}

• Setting z = e^{jω} means r = 1 — walking around the unit circle in the z-plane traces out the DTFT.

• This only works if the unit circle lies inside the Z-transform's region of convergence — otherwise the DTFT simply doesn't exist for that
sequence.

• This is exactly why the Region of Convergence (next) is not a footnote — it determines whether, and how, a sequence has a valid
frequency-domain description.

Session 1 · Ch. 3.1 DSP · Week 4

 The Region of Convergence (ROC)
The set of z for which the Z-transform sum actually converges

• X(z) is only meaningful together with its ROC — the same algebraic expression for X(z) can correspond to different sequences x[n],
depending on the ROC.

• The ROC is always a ring (annulus) centered on the origin of the z-plane: r₁ < |z| < r₂ — it depends only on |z|, never on the angle ω.

• The ROC can never contain a pole of X(z), since X(z) is infinite there.

• Its shape (disk exterior, disk interior, or annulus) depends on whether the sequence is right-sided, left-sided, or two-sided — shown
next.

Session 1 · Ch. 3.2 DSP · Week 4

 ROC Shapes by Sequence Type
The ROC's geometry tells you immediately what kind of sequence you have

Right-sided (causal) Left-sided (anti-causal) Two-sided

ROC: |z| > r ROC: |z| < r ROC: r₁ < |z| < r₂

Shaded region = ROC. Red × marks a pole (always excluded from the ROC).
Session 1 · Ch. 3.2 DSP · Week 4
 Worked Example: Causal Exponential
x[n] = aⁿ·u[n]

X(z) = Σₙ₌₀∞ aⁿ z⁻ⁿ = Σₙ₌₀∞ (a z⁻¹)ⁿ

This is a geometric series, which converges iff |a z⁻¹| < 1, i.e. |z| > |a|

X(z) = 1 / (1 − a z⁻¹), ROC: |z| > |a|

A single pole at z = a. If |a| < 1 the unit circle lies inside the ROC, so the DTFT exists and equals X(z) at z = e^{jω} — matching our Week 2 DTFT
example.

Session 1 · Ch. 3.2 DSP · Week 4

Session 1 — Key Takeaways

• X(z) = Σₙ x[n]·z⁻ⁿ generalizes the DTFT by adding a magnitude degree of freedom, r = |z|

• The DTFT is X(z) restricted to the unit circle, |z| = 1 — valid only if the unit circle lies in the ROC

• The ROC is always an annulus in |z|, and it never contains a pole

• Right-sided sequences → ROC exterior to a circle; left-sided → interior; two-sided → an annulus

Maps to CLO1 · Reading: Oppenheim & Schafer, Ch. 3.1–3.2

SESSION 2

Z-Transform Properties & Inverse Z-

Transform
Poles, zeros, common pairs, and getting back to x[n] · Ch. 3.3–3.4
 Z-Transform Properties at a Glance
The algebraic toolkit for manipulating and solving with X(z)

Property Time domain z-domain

Linearity a·x₁[n] + b·x₂[n] a·X₁(z) + b·X₂(z)

Time shift x[n−n₀] z⁻ⁿ⁰·X(z)

Scaling in z aⁿ·x[n] X(z/a)

Time reversal x[−n] X(1/z)

Convolution x[n] * h[n] X(z)·H(z)

Differentiation n·x[n] −z·dX(z)/dz

Session 2 · Ch. 3.3 DSP · Week 4

 Poles, Zeros, and Pole-Zero Plots
X(z) written as a ratio of polynomials is fully described by its poles and zeros

X(z) = N(z) / D(z) — a rational function of z

• Zeros: values of z where N(z) = 0 (X(z) = 0)
• Poles: values of z where D(z) = 0 (X(z) → ∞)
• The ROC is bounded by pole locations — it's always a ring or disk whose edges pass
through poles, never through a pole itself.
• Complex poles/zeros of a real sequence always occur in conjugate pairs.

× = pole, ○ = zero

Session 2 · Ch. 3.3 DSP · Week 4

 Common Z-Transform Pairs
Worth memorizing — these cover most of the sequences you'll encounter

x[n] X(z) ROC

δ[n] 1 all z

u[n] 1 / (1 − z⁻¹) |z| > 1

aⁿ u[n] 1 / (1 − a z⁻¹) |z| > |a|

−aⁿ u[−n−1] 1 / (1 − a z⁻¹) |z| < |a|

n aⁿ u[n] a z⁻¹ / (1 − a z⁻¹)² |z| > |a|

Session 2 · Ch. 3.3 DSP · Week 4

 Getting Back to x[n]: The Inverse Z-Transform
Three practical methods, depending on the form of X(z)

Contour Integration Partial Fraction Expansion Power Series (Long Division)

The formal definition, via a contour integral Break a rational X(z) into simple terms Expand X(z) directly as a power series in z⁻¹
in the ROC. Rarely used by hand — mainly matching known Z-transform pairs. The — useful when you just need the first
a theoretical foundation. workhorse method for LTI system analysis. several values of x[n].

Session 2 · Ch. 3.4 DSP · Week 4

 Worked Example: Partial Fraction Expansion
X(z) = 1 / [(1 − 0.5z⁻¹)(1 − 0.25z⁻¹)], ROC: |z| > 0.5

Step 1 — Expand into simple fractions:

X(z) = A / (1 − 0.5z⁻¹) + B / (1 − 0.25z⁻¹)

Step 2 — Solve for A and B (cover-up method or equating coefficients):

A = 2, B = −1

Step 3 — Match each term to a known pair (ROC tells us both are right-sided):

x[n] = [ 2·(0.5)ⁿ − (0.25)ⁿ ]·u[n]

Session 2 · Ch. 3.4 DSP · Week 4

 Power Series Method (Long Division)
Directly reading off x[n] as the coefficients of z⁻ⁿ

• Since X(z) = Σₙ x[n]·z⁻ⁿ, if you can expand X(z) as a power series in z⁻¹, the coefficients are the sequence values directly.

• For a rational X(z), this means performing polynomial long division of the numerator by the denominator (in ascending powers of
z⁻¹ for a causal / right-sided expansion).

• Very useful for a quick numerical check (first few samples) or for sequences with no simple closed form, but tedious for finding a
general formula for x[n].

Session 2 · Ch. 3.4 DSP · Week 4

Session 2 — Key Takeaways

• Z-transform properties (linearity, shift, scaling, convolution) mirror the DTFT's, with the ROC tracked alongside every
operation

• A rational X(z) is fully characterized by its poles, zeros, and ROC

• Partial fraction expansion is the standard method for inverting a rational X(z) by hand

• The ROC determines which inverse (causal, anti-causal, or two-sided) matches a given X(z)

Maps to CLO1 · Reading: Oppenheim & Schafer, Ch. 3.3–3.4

Week 4 Recap

• The Z-transform X(z) = Σₙ x[n]z⁻ⁿ generalizes the DTFT, with the ROC as an essential companion, not an afterthought
• Right-/left-/two-sided sequences map to exterior/interior/annular ROCs
• Poles and zeros fully characterize a rational X(z); partial fraction expansion is the main tool for inverting it

Next Week: Z-Transform Analysis of LTI Systems

Stability, causality, and frequency response directly from pole-zero plots (Ch. 3.5–3.6)

CLO1 · Oppenheim & Schafer, Ch. 3.1–3.4