Evolution of the Z-Transform

1. The Problem Domain: Difference Equations

Before any transform method is invented, there must be a problem to solve. For the z-transform, that problem was the analysis of linear, time-invariant (LTI) discrete-time systems, which are described by linear constant-coefficient difference equations.

You've encountered linear constant-coefficient differential equations extensively in your study of analog circuits and control systems. A difference equation is its discrete-time counterpart. Instead of relating a function and its derivatives, it relates terms in a sequence at different time indices.

For example, an equation like $y[n]=0.5y[n-1]+x[n]$ describes a simple discrete-time system where the current output $y[n]$ depends on the previous output $y[n-1]$ and the current input $x[n]$.

To get a feel for what these are and how they were solved "the hard way" before transform methods became standard, let's watch a brief video.

As the video demonstrates, you can solve these equations by finding a homogeneous and a particular solution, much like with differential equations. However, this process can become tedious, especially for high-order systems or complex inputs. This created a clear need for a more powerful, systematic approach—a transform that could convert difference equations into simpler algebraic equations.

2. A Historical Journey to the z-Transform

The idea of using a transform to solve such problems didn't appear overnight. It was the culmination of centuries of mathematical thought and decades of engineering necessity.

Let's begin by reading a concise history of the z-transform's development.

Let's piece that timeline together and add some context.

The Mathematical Foundation (pre-1940s)

As the article mentions, the core idea is not new.

• Generating Functions (de Moivre, 1730s): In mathematics, a generating function is a way to encode an infinite sequence of numbers $(a_0,a_1,a_2,\dots )$ as a single formal power series, $G(x)={\sum }_{n=0}^{\infty }{a}_n{x}^n$. Manipulating the function $G(x)$ can reveal properties of the sequence $a_n$. You can immediately see the structural similarity to the eventual z-transform.
• Laplace (late 1700s): Pierre-Simon Laplace's work, which led to the Laplace transform, also involved these ideas. His transform deals with continuous functions, but the underlying principle of turning a complex problem (calculus) into a simpler one (algebra) is the same.
• Laurent Series (1843): Mathematically, the z-transform is a specific type of Laurent series, which is a power series that can include terms with negative powers. This is the formal framework that allows the transform to handle sequences that exist for both positive and negative time indices.

The Engineering Need (1940s)

The theoretical tools existed, but the impetus to apply them to engineering came with the advent of sampled-data systems. During and after World War II, technologies like radar required processing signals that were discrete in time (sampled). Engineers needed a systematic way to analyze these systems.

Early attempts, mentioned in one of our resources, involved using the Laplace transform in complex ways to handle difference equations, but these methods were cumbersome. A more direct approach was needed.

The Birth of the Modern Transform (1947-1952)

This is where the key players enter the scene.

• Witold Hurewicz (1947): While working on servo-mechanisms, Hurewicz developed a method for solving linear difference equations by defining a transform of the sampled signal. This was a crucial step toward a dedicated discrete-time transform.

• Ragazzini and Zadeh (1952): At Columbia University, a group working on sampled-data control systems, led by John Ragazzini and including a young Lotfi Zadeh, refined and unified these emerging ideas. In a landmark 1952 paper, they formally introduced what they called "the z-transform". The name simply came from the complex variable, $z$, used in the definition.

3. The Modern Formulation

This historical journey leads us to the z-transform as we know it today. It is defined as a power series where the terms of our discrete sequence $x[n]$ serve as the coefficients.

The bilateral (or two-sided) z-transform of a discrete-time signal $x[n]$ is defined as:

where $z$ is a complex variable.

The video we started with provides a concise presentation of this modern definition and its purpose.

The process, which we will explore in detail in later lessons, is:

1. Take the z-transform of the entire difference equation.
2. Solve the resulting algebraic equation for $Y(z)$, the transform of the output.
3. Take the inverse z-transform to find the output signal $y[n]$.

This is a direct parallel to using the Laplace transform to solve differential equations.

4. Connecting the s-Plane and the z-Plane

Perhaps the most elegant part of this story for an Electrical Engineer is the direct mathematical link between the Laplace transform and the z-transform. The z-transform isn't just an analogous tool; it's a direct consequence of sampling a continuous-time signal.

Recall that a sampled signal can be modeled as the original continuous signal $x(t)$ multiplied by a train of Dirac impulses spaced by the sampling period $T_s$. If we take the Laplace transform of this sampled signal, we arrive at an expression that is equivalent to the z-transform of the discrete sequence $x[n]=x(n{T}_s)$ under the substitution:

where $s$ is the complex variable from the Laplace domain.

This single equation is the Rosetta Stone connecting the continuous and discrete transform worlds. It explains, for instance, why stability in the s-plane (requiring poles in the left-half plane, $\text{Re}(s)<0$) corresponds to stability in the z-plane (requiring poles inside the unit circle, $|z|<1$).

Conclusion

In this lesson, we traced the evolution of the z-transform. We saw that it wasn't an arbitrary invention but a tool developed to meet a specific engineering need—the analysis of sampled-data systems described by difference equations. Its formulation draws on deep mathematical concepts and is elegantly connected to the Laplace transform you already know.

Key Takeaways:

• The z-transform was developed to be a transform method for solving linear constant-coefficient difference equations, the discrete-time equivalent of differential equations.
• Its mathematical ideas can be traced back to generating functions (de Moivre, 1730s) and are formally described by Laurent series.
• The practical need arose in the 1940s with the growth of sampled-data control systems (e.g., radar).
• Key historical figures include Witold Hurewicz (1947), who developed an early version, and John Ragazzini and Lotfi Zadeh (1952), who unified the approach and coined the name "z-transform".
• The z-transform is formally connected to the Laplace transform through the mapping $z=e^{s{T}_s}$, which links the s-plane to the z-plane.

Preview of the Next Lesson:

Now that we understand the origin and purpose of the z-transform, we'll begin our technical exploration. In the next lesson, we will derive the Discrete-Time Fourier Transform (DTFT) and establish its relationship to the continuous-time Fourier transform of a sampled signal. We will see that the DTFT is, in fact, a special case of the z-transform evaluated on the unit circle ($|z|=1$), which is the discrete-time parallel to the Fourier transform being the Laplace transform evaluated on the imaginary axis ($s=j\Omega$).