⚡ Spark Academy53 lessons

Fourier: Thinking in Frequencies

Every repeating wave is secretly a chord of pure sines. Learn to read the recipe and half of engineering becomes legible.

lesson 2 of 3 in this unit

Builds on: 5.1 Alternating Current5.3 Impedance & RC Filters13.1 Sampling & Aliasing

The boldest claim in mathematics-for-engineers

In 1807 Joseph Fourier claimed that any repeating waveform — square, sawtooth, a violin note, your voice — is a sum of pure sine waves: a fundamental at the repeat rate plus harmonics at exact integer multiples. The committee reviewing the paper (including Lagrange) didn’t fully believe him. He was right, and the consequences run every technology that touches a signal.

wave = a₁·sin(2πf·t) + a₂·sin(2π·2f·t) + a₃·sin(2π·3f·t) + …the list of amplitudes (a₁, a₂, a₃ …) is the wave's spectrum — its recipe

The classic recipes are worth recognising on sight: a square wave is odd harmonics fading as 1/k (1, ⅓, ⅕, …); a triangle is odd harmonics fading much faster (1/k²) — which is why it sounds mellow and the square sounds buzzy; a sawtooth contains every harmonic — the brashest of all, and the reason it powers synthesizer basses.

Two views, one signal

The oscilloscope view (voltage vs time) and the spectrum view (amplitude vs frequency) are the same information displayed differently — and many problems are only easy in one of them. Why does a low-pass filter turn a square wave into a sine? Time view: mysterious rounding. Frequency view: obvious — the filter ate the harmonics and left the fundamental. Why does your 1.4 kHz PWM need filtering to become clean DC (10.2)? Its spectrum is a DC term plus harmonics of 1.4 kHz; the LC filter keeps the DC and dumps the rest.

Sharp edges are the giveaway: fast transitions demand high harmonics. A crisp square wave needs frequency content far above its repeat rate — which is why digital signals radiate interference (every edge is a little broadcast), and why Nyquist’s fmax in the last lesson means the highest harmonic, not the repeat rate.

The FFT

Computers extract spectra with the Fast Fourier Transform — arguably the most important algorithm ever written. Your phone runs thousands per second: Wi-Fi, voice calls, music apps that show you a spectrum analyser. Every one is Fourier’s 1807 idea, at gigahertz.

⚡ Lab — The Wave Kitchen

Build the classic waveforms one sine at a time, recipe on the right.

  • Square with 1 harmonic: just a sine. With 25: crisp shoulders. Watch it converge.
  • Switch to triangle — notice how few harmonics it needs (1/k² dies fast).
  • Sawtooth: every bar in the spectrum is lit. Hence the buzz.
3
Fourier's claim
any repeating wave = a fundamental sine plus harmonics at integer multiples — no exceptions
Why filters care
a 100 Hz square wave has energy at 300, 500, 700 Hz… — a low-pass that eats the harmonics rounds it into a sine

Check your understanding

Q1. A 100 Hz square wave contains energy at…

Q2. A low-pass filter turns a square wave into a near-sine because…

Q3. Sharp, fast edges in a waveform imply…

Q4. The spectrum of a signal is…