Fourier analysis is a way to study complicated signals by breaking them into simpler sine and cosine waves. This cheat sheet helps students connect equations to real applications such as sound, images, circuits, and data filtering. It is useful because many real-world patterns are easier to understand in the frequency domain than in the time domain.
The main idea is that a signal can be represented as a sum or spread of waves with different frequencies, amplitudes, and phases. Fourier series are used for periodic signals, while Fourier transforms are used for nonperiodic or continuous signals. Important applications include identifying dominant frequencies, removing noise, compressing data, and solving differential equations.
Key Facts
- A periodic function f(t) with period T can be written as a Fourier series: f(t) = a0/2 + sum from n = 1 to infinity of [an cos(nω0t) + bn sin(nω0t)], where ω0 = 2π/T.
- The Fourier transform changes a time-domain signal into a frequency-domain function: F(ω) = integral from negative infinity to infinity of f(t)e^(-iωt) dt.
- The inverse Fourier transform reconstructs the original signal: f(t) = 1/(2π) integral from negative infinity to infinity of F(ω)e^(iωt) dω.
- Amplitude tells how strong a frequency component is, while phase tells how that component is shifted in time.
- A low-pass filter keeps low frequencies and reduces high frequencies, which is useful for smoothing data or removing sharp noise.
- A high-pass filter keeps high frequencies and reduces low frequencies, which is useful for detecting edges in images or fast changes in signals.
- The sampling rate must be at least twice the highest frequency in a signal to avoid aliasing, so fs >= 2fmax.
- Convolution in the time domain becomes multiplication in the frequency domain: Fourier transform of (f * g) = F(ω)G(ω).
Vocabulary
- Fourier Series
- A representation of a periodic function as a sum of sine and cosine waves.
- Fourier Transform
- A mathematical operation that converts a signal from time or space information into frequency information.
- Frequency Spectrum
- A display or function showing which frequencies are present in a signal and how strong they are.
- Amplitude
- The size or strength of a wave component in a signal.
- Phase
- The horizontal shift of a wave component relative to a reference wave.
- Aliasing
- A sampling error that makes high-frequency signals appear as false lower-frequency signals.
Common Mistakes to Avoid
- Confusing time domain and frequency domain is wrong because the time domain shows how a signal changes over time, while the frequency domain shows which wave components make it up.
- Using frequency f and angular frequency ω as if they are the same is wrong because they differ by the formula ω = 2πf.
- Ignoring phase is wrong because two signals can have the same amplitudes at each frequency but still look different if their phases are different.
- Sampling below the Nyquist rate is wrong because frequencies above half the sampling rate can be misread as lower frequencies.
- Assuming filtering only changes noise is wrong because a filter can also remove or weaken useful parts of the original signal if their frequencies overlap with the noise.
Practice Questions
- 1 A sound wave has period T = 0.005 s. Find its fundamental frequency f0 and angular frequency ω0.
- 2 A sensor samples data at 200 Hz. What is the highest frequency that can be measured without aliasing?
- 3 A signal is f(t) = 3 cos(10t) + 2 sin(30t). Identify the angular frequencies present and the amplitude of each component.
- 4 Explain why Fourier analysis is useful for removing high-frequency noise from a slowly changing temperature signal.
Understanding Fourier Analysis Applications
A useful way to picture Fourier analysis is as a recipe for rebuilding a waveform. Each ingredient has a size, a rate of repetition, and a timing shift. A square-looking electrical signal, for example, needs many higher frequency ingredients because its sharp corners change very quickly.
A smooth signal needs fewer strong high frequency parts. This explains why frequency content is closely linked to detail.
Fast changes in time create broad frequency content. Slow changes produce energy near low frequencies.
Digital devices do not work with infinitely long, perfectly continuous signals. They record a finite list of measurements. In practice, students often use a discrete Fourier transform, usually computed by a fast Fourier transform algorithm.
This produces frequency bins. Each bin represents a small range of frequencies rather than one perfectly exact frequency. The length of the recording matters.
A longer recording gives narrower bins and makes nearby tones easier to separate. If a recorded tone does not fit neatly into the measured time interval, its energy can spread into nearby bins.
This effect is called spectral leakage. Window functions reduce leakage by gently lowering the signal near the start and end of the record.
Sampling needs careful attention before any transform is calculated. A microphone, camera sensor, or motion tracker takes readings at regular intervals. Frequencies above half the sampling rate can appear as false lower frequencies in the data.
This is aliasing. It cannot be fixed reliably after recording because different original signals can create the same sampled data. Engineers place an analog low-pass filter before sampling to remove frequencies that the device cannot measure safely.
In music production, an alias can sound like an unwanted low tone. In video, it can create flickering patterns or false stripes on fine detail.
Filtering is not magic removal of unwanted parts. A filter changes the signal, so it always involves a tradeoff. Strong smoothing can reduce random measurement noise, yet it can blur a sudden event that matters.
A high-pass filter can reveal a heartbeat change or an image boundary, yet it may amplify noisy fluctuations. Filter design considers the cutoff frequency, how quickly the filter changes between kept and reduced frequencies, and whether the output is delayed or distorted around sharp events. The convolution rule makes filtering efficient because a filter can be applied by multiplying frequency components.
When learning this topic, inspect the original signal and the reconstructed result together. Check units, sampling rate, recording length, and phase, not just the tallest peaks in a spectrum.