1 / 14
Chapter 3 · Laplace Transforms

Why Laplace Transforms Exist

The idea before the machinery
Dr. Mohamed Mabrok · Qatar University

What Chapter 2 Costs You

The ritual
1. Solve the homogeneous equation
2. Guess the shape of $y_p$
3. Substitute, match coefficients
4. Only now use $y(0)$ and $y'(0)$
The complaint
The two pieces of physical information you actually started with — the initial conditions — arrive last. They are an afterthought.
And then it breaks
Switch the voltage on after two seconds. Strike the system once with a hammer. Now try to guess a particular solution.

Where the Method Falls Apart

t t = 2 switched on t one sharp blow
The honest version of engineering
Real systems are switched on, switched off, and hit. Undetermined coefficients has nothing to guess here — and this is exactly where the subject gets interesting.

You Have Done This Before: Logarithms

Before calculators
To multiply two six-digit numbers, nobody multiplied.
$$a \times b \;\longrightarrow\; \log a + \log b \;\longrightarrow\; \text{antilog}$$
A hard operation became an easy one, in a different world.
The same move, for calculus
A problem about derivatives becomes a problem about algebra.
$$\frac{d}{dt} \;\longrightarrow\; \text{multiply by } s$$
Solve the easy thing. Then come back.
$$\text{Transform} \;\longrightarrow\; \text{Solve (algebra)} \;\longrightarrow\; \text{Transform back}$$
That is the whole chapter. Everything else is detail.

The Definition — Read It As A Sentence

$$\mathcal{L}\{f(t)\} = F(s) = \int_0^{\infty} e^{-st} f(t)\,dt$$
Take your signal $f(t)$ — the whole of it, forward in time.
Weight it, at every instant, by the fading curve $e^{-st}$.
Add up what is left. The integral is the adding up.
Repeat for every $s$. Those numbers are the function $F(s)$.
What just happened
The entire history of a signal was crushed into one number — once for each value of $s$.

The Weight Is A Spotlight That Dims

t the weight $e^{-st}$ fades your signal f(t)
Read it this way
Early time is examined under a bright light. Late time is examined under a dim one. How quickly the light dims is set entirely by $s$.

So What Does $s$ Actually Control?

Large $s$
t
The light dies at once. Only the first instants are measured.
Medium $s$
t
A balance: the near future carries most of the weight.
Small $s$
t
The light lingers. The long run counts almost as much as now.
The picture to keep
$F(s)$ is your signal examined at every time scale at once. A prism spreads light into its colours; this spreads a signal into its rates of growth and decay.

Test The Reading On Two Examples

The constant
$$\mathcal{L}\{1\} = \frac{1}{s}, \qquad s \gt 0$$
Large $s$: the light died early, little was captured, $1/s$ is small.
Small $s$: the light lingered, much was captured, $1/s$ is large.
At $s = 0$ the light never fades — you are adding a constant forever. Infinite. That is where the condition comes from.
The exponential
$$\mathcal{L}\{e^{at}\} = \frac{1}{s-a}, \qquad s \gt a$$
The signal grows at rate $a$. The light fades at rate $s$. The condition $s \gt a$ is simply: the fade must beat the growth.
These conditions are not bureaucracy. They are the honest limit of what can be summed.

Read It As A Race

t signal grows: e^(at) light fades: e^(-st) their product — settles, so the integral exists
Fade wins ($s \gt a$)
The product dies away. The area is finite. You get a transform.
Growth wins ($s \lt a$)
The product runs away. There is nothing to add up. No transform.

And At The Edge Of The Race: A Pole

s s = a F(s) = 1/(s − a) the transform blows up exactly at the signal's own rate
Remember this one
A pole sits precisely at the natural rate of the system. The transform carries a fingerprint, and the fingerprint points at the behaviour the system wants to have on its own.
In Chapter 4 this is what you will read off the page: where the poles sit decides whether a design settles or shakes itself apart.

The Property That Makes It A Method

$$\mathcal{L}\{f'(t)\} = sF(s) - f(0)$$
The derivative is gone
On the left, calculus. On the right, multiplication by $s$. This is the logarithm promise, delivered.
The initial condition walked in
$f(0)$ arrived by itself — it is the boundary term from integration by parts, and the boundary of this integral is $t = 0$.
Why that matters
The state you start from is built into the method from the first line — not imposed at the end, as it was in Chapter 2.

Watch The Loop Close

$$y' + 3y = 0, \qquad y(0) = 2$$
1 Transform: $\;sY - 2 + 3Y = 0$  — no derivatives anywhere.
2 Solve, in one line: $\;Y = \dfrac{2}{s+3}$
3 Read it back: a pole at $-3$, so decay at rate 3 — $\;y = 2e^{-3t}$
$$\text{ODE} \;\xrightarrow{\;\mathcal{L}\;}\; \text{algebra in } s \;\xrightarrow{\;\text{solve}\;}\; Y(s) \;\xrightarrow{\;\mathcal{L}^{-1}\;}\; y(t)$$
The differential equation never had to be treated as a differential equation at all.

What It Costs You

It starts at zero
The integral runs from $0$ to $\infty$, so the transform is built for systems that start: switched on, released from rest, struck at a known moment.
For an engineer that is not a weakness. It is almost every problem you will be handed.
It must be out-run
If a function grows faster than every exponential, no $s$ can tame it and it has no transform.
Such functions are rare, and none of them appear in this course.
And coming back?
The transform table is your table of logarithms, read in the other direction. Partial fractions are how you get your answer into a shape the table recognises.

If You Remember One Sentence

Differentiation becomes multiplication by $s$.
The time world
A system is described by derivatives. Hard.
The $s$ world
The same system is described by multiplication. Easy.
Next
The table, the shifting theorems for systems switched on late, and the derivative rule applied twice — where both initial conditions walk in by themselves.
🔊 Voice
1×