Fourier series of periodic signals

Represents periodic signals as sums of harmonics in trigonometric and exponential form, with symmetry shortcuts, convergence, Gibbs phenomenon and Parseval's power relation.

Drafted with Aria, reviewed by the AiCanCode.org team. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Mains-driven instruments, PWM drives, rotating machinery and clock lines all produce periodic signals, and their behaviour is described by harmonics. The Fourier series tells you how much of a periodic signal sits at each multiple of the fundamental frequency, which is exactly what a harmonic analyser, a THD meter or a filter design needs.

Key ideas

The idea. A periodic signal x(t) with period T0 can be written as a sum of sinusoids at the fundamental frequency f0 = 1/T0 and its integer multiples (harmonics). There is no energy at any other frequency: the spectrum is a set of lines at k·f0.

Two equivalent forms

  • Trigonometric: DC term a0 plus ak cos(kω0t) + bk sin(kω0t).
  • Complex exponential: Σ ck e^(jkω0t) for k from −∞ to ∞. For a real signal, c(−k) = ck*, so the two-sided magnitude spectrum is even and the phase spectrum is odd.
  • Polar (amplitude–phase) form: a0 + Σ Ak cos(kω0t + φk), with Ak = 2|ck|.

Dirichlet conditions (sufficient for convergence): over one period x(t) is absolutely integrable and has a finite number of maxima, minima and finite discontinuities. At a point of continuity the series converges to x(t); at a jump it converges to the mid-point of the jump.

Gibbs phenomenon. A truncated series overshoots near a discontinuity by about 9 % of the jump height. Adding more terms makes the ripple narrower but does not reduce the peak overshoot.

Symmetry shortcuts (they save most of the integration in exams)

  • Even x(−t) = x(t): bk = 0 — cosine terms only (plus DC).
  • Odd x(−t) = −x(t): a0 = 0 and ak = 0 — sine terms only.
  • Half-wave symmetry x(t ± T0/2) = −x(t): only odd harmonics are present, and a0 = 0.
  • Removing a DC offset changes only a0; shifting in time changes only the phases.

Decay of coefficients. Coefficients of a signal with jumps (square wave) fall as 1/k; a continuous signal with slope jumps (triangle wave) falls as 1/k². Smoother signals have fewer significant harmonics.

Parseval's relation. The average power equals the sum of the powers in each harmonic — the basis of THD calculations.

Properties (complex form, x ↔ ck): linearity; time shift x(t − t0) ↔ ck e^(−jkω0t0) (magnitudes unchanged); differentiation dx/dt ↔ jkω0 ck; time reversal x(−t) ↔ c(−k).

LTI response. If a periodic x(t) drives an LTI system with frequency response H(jω), the output is periodic with coefficients ck·H(jkω0) — each harmonic is scaled and phase-shifted separately.

Formulas

  • x(t) = a0 + Σ from k=1 to ∞ of [ak cos(kω0t) + bk sin(kω0t)] — trigonometric series.
  • a0 = (1/T0) ∫ over T0 of x(t) dt — DC (average) value.
  • ak = (2/T0) ∫ over T0 of x(t) cos(kω0t) dt, bk = (2/T0) ∫ over T0 of x(t) sin(kω0t) dt.
  • x(t) = Σ ck e^(jkω0t), ck = (1/T0) ∫ over T0 of x(t) e^(−jkω0t) dt — exponential series.
  • c0 = a0, ck = (ak − j·bk)/2 for k ≥ 1.
  • P = (1/T0) ∫ over T0 of |x(t)|² dt = Σ |ck|² = a0² + Σ (ak² + bk²)/2 — Parseval.
  • x(t) = (4A/π) Σ over odd k of sin(kω0t)/k — odd square wave between +A and −A.
  • ck = (A·τ/T0) · sinc(kτ/T0), with sinc(u) = sin(πu)/(πu) — rectangular pulse train of height A and width τ, centred at t = 0.
  • THD = √(P − P1)/√P1 (powers of the AC signal, DC excluded) — total harmonic distortion.

Symbols: T0 fundamental period (s); f0 = 1/T0 (Hz); ω0 = 2π/T0 (rad/s); k harmonic number; a0, ak, bk, ck, A in the unit of x (V, A, …); τ pulse width (s); P normalised power (unit of x)²; P1 power of the fundamental.

Worked examples

Example 1 (standard). A ±5 V square wave with period 2 ms is odd about t = 0. Find the amplitudes of the fundamental and third harmonic, and the fraction of the power in the fundamental.

  1. Odd and half-wave symmetric ⇒ only bk, odd k: bk = 4A/(kπ).
  2. f0 = 1/(2 ms) = 500 Hz. b1 = 4 × 5/π = 6.366 V at 500 Hz.
  3. b3 = 4 × 5/(3π) = 2.122 V at 1.5 kHz.
  4. Total power P = A² = 25 V² (the square of ±5 V is always 25).
  5. P1 = b1²/2 = 6.366²/2 = 20.26 V², so P1/P = 8/π² = 0.811.
  6. b1 ≈ 6.37 V, b3 ≈ 2.12 V; about 81.1 % of the power is in the fundamental (THD = √(1 − 0.811)/√0.811 ≈ 48.3 %).

Example 2 (GATE level). A pulse train has height 1, width τ = T0/4, centred at t = 0. Find c0, c1, c2, c4, and the fraction of the total power contained in the DC term plus the fundamental.

  1. ck = (τ/T0) sinc(kτ/T0) = (1/4) · sin(kπ/4)/(kπ/4) = sin(kπ/4)/(kπ).
  2. c0 = τ/T0 = 0.25; c1 = sin(π/4)/π = 0.2251; c2 = sin(π/2)/(2π) = 0.1592; c4 = sin(π)/(4π) = 0 (every 4th harmonic is missing).
  3. Total power P = (1/T0) ∫ x² dt = τ/T0 = 0.25.
  4. DC + fundamental (both k = ±1): c0² + 2|c1|² = 0.0625 + 2 × 0.05066 = 0.1638.
  5. Fraction = 0.1638/0.25. About 65.5 % of the power is in DC plus the fundamental.

Common mistakes

  • Using 2/T0 for a0. The DC term is the average, 1/T0. (Some books write the series with a0/2; then a0 = (2/T0)∫x dt. Check which convention the question uses.)
  • Forgetting the negative-frequency coefficients in Parseval's sum: a harmonic contributes 2|ck|².
  • Claiming that more terms remove the Gibbs overshoot.
  • Expecting a square wave to contain even harmonics — with half-wave symmetry it has none.
  • Writing the period of cos(3πt) + sin(6πt) as anything but 2/3 s; ω0 = 3π, so the 6π term is the second harmonic.
  • Mixing ck (two-sided, half amplitude) with Ak (one-sided amplitude) when reading a spectrum.

For GATE IN

Typical questions: identify which harmonics or coefficients are zero from symmetry; compute a0, a specific ck or bk; find total or fractional power with Parseval; find the output of an LTI system (often an ideal filter) for a periodic input by keeping only the harmonics inside its passband. Practise the standard pairs (square, triangle, pulse train, half- and full-wave rectified sine) and the sinc envelope of a pulse train.

Quick check

  1. Which coefficients vanish for an even signal?
  2. What value does the series converge to at a jump from −1 to +3?
  3. A periodic signal has c0 = 1, c±1 = 0.5, all others zero. What is its power?
  4. A full-wave rectified sine contains which harmonics of the mains frequency?
  5. What is a0 of x(t) = 4 + 3 cos(2πt)? Answers: 1. All bk. 2. The mid-point, 1. 3. 1 + 0.25 + 0.25 = 1.5. 4. DC and even harmonics only (its own fundamental is twice the mains frequency). 5. 4.

Try answering each one aloud before you open it.

  1. 1.What is a Fourier series and why is it important in analyzing periodic signals?Concept

    A Fourier series is a way to represent a periodic signal as a sum of simple sine and cosine waves. It is important because it allows us to analyze and understand the frequency components of a signal, which is crucial in fields like signal processing and communications.

  2. 2.Explain the difference between the Fourier series and the Fourier transform.Concept

    The Fourier series is used to represent periodic signals as a sum of sine and cosine functions, while the Fourier transform is used for non-periodic signals to transform them into the frequency domain. The Fourier series results in discrete frequency components, whereas the Fourier transform results in a continuous spectrum.

  3. 3.How do you determine the coefficients of a Fourier series for a given periodic signal?Concept

    The coefficients of a Fourier series are determined using integrals. For a periodic function f(t) with period T, the coefficients a₀, aₙ, and bₙ are calculated using specific integral formulas over one period of the function. These coefficients represent the amplitude of the corresponding sine and cosine terms.

  4. 4.Why are Fourier series used in signal processing?Application

    Fourier series are used in signal processing because they allow us to decompose complex periodic signals into simpler sinusoidal components. This decomposition helps in analyzing the frequency content of signals, filtering, and reconstructing signals, which are essential tasks in signal processing.

  5. 5.What happens if a signal is not periodic? Can it still be analyzed using Fourier series?Application

    If a signal is not periodic, it cannot be directly analyzed using Fourier series. However, it can be approximated as periodic over a large interval or analyzed using the Fourier transform, which is suitable for non-periodic signals.

  6. 6.Explain how the Gibbs phenomenon affects the Fourier series representation of a signal.Concept

    The Gibbs phenomenon refers to the overshoot (or 'ringing') that occurs near discontinuities in the Fourier series approximation of a signal. This overshoot does not disappear as more terms are added to the series, but its width decreases. It highlights the limitations of Fourier series in perfectly reconstructing signals with sharp transitions.

  7. 7.Why is it important to consider the convergence of a Fourier series?Application

    Convergence of a Fourier series is important because it determines whether the series accurately represents the original signal. If a Fourier series converges, it means that as more terms are added, the series approaches the actual signal. Non-convergence can lead to incorrect signal representation.

  8. 8.What are the Fourier series coefficients of a square wave that switches between +A and −A and is odd about t = 0?Numerical

    The wave is odd, so a0 = 0 and all ak = 0; it also has half-wave symmetry, so only odd harmonics appear. The sine coefficients are bk = 4A/(kπ) for odd k and 0 for even k, giving x(t) = (4A/π)[sin ω0t + (1/3)sin 3ω0t + (1/5)sin 5ω0t + …]. The 1/k decay reflects the jumps in the waveform, and about 81% (8/π²) of the power is in the fundamental.

  9. 9.What is the effect of truncating a Fourier series after a finite number of terms?Application

    Truncating a Fourier series after a finite number of terms results in an approximation of the original signal. The more terms included, the closer the approximation is to the actual signal. However, truncation can lead to phenomena like Gibbs phenomenon, especially near discontinuities.

  10. 10.What is the Fourier series of a triangular wave that swings between +A and −A and is even about t = 0 (peak at t = 0)?Numerical

    The wave is even and half-wave symmetric, so it has only cosine terms at odd harmonics and no DC: x(t) = (8A/π²) Σ over odd k of cos(kω0t)/k². The coefficients fall as 1/k² because the triangle is continuous and only its slope jumps, so it converges much faster than a square wave and shows no Gibbs overshoot.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?