Physics
Power Spectral Density
Quick fact
Power spectral density is measured in units of power per frequency (e.g., W/Hz) and is particularly useful for analyzing random or continuous signals where the total power is spread over a range of frequencies.
Why this is interesting
You hear a complex sound—like a bustling street. How can you tell which frequencies carry the most energy? Power spectral density reveals exactly that.
Read the full explanation
Understanding Power Spectral Density
Imagine a prism splitting white light into a rainbow: each color corresponds to a specific range of frequencies, and the intensity of each color indicates how much energy is present. Power spectral density does the same for any signal—whether it's sound, radio waves, or vibrations. Instead of plotting the signal's amplitude over time (time domain), we plot how its power is distributed across frequencies (frequency domain). For a pure tone (like a single musical note), the PSD shows a sharp peak at that tone's frequency. For noise, it can be flat (white noise) or shaped (colored noise). Understanding PSD helps engineers design filters, diagnose machinery faults from vibration signatures, and characterize wireless communication channels.
A deeper explanation
Mathematically, the power spectral density (PSD) of a stationary random process is defined as the Fourier transform of its autocorrelation function. The autocorrelation measures how the signal is correlated with itself at different time lags; its Fourier transform then reveals the frequency content. Equivalently, for deterministic signals, the PSD is the squared magnitude of the Fourier transform normalized by the signal length. This representation is crucial because many natural and engineered signals are not perfectly periodic—they contain randomness (e.g., thermal noise, turbulent wind). PSD allows us to quantify the average power in each frequency band, even for such signals. Key applications include identifying resonant frequencies in structures, measuring noise floors in electronics, and analyzing brain wave activity (EEG). The concept matters because it provides a compact yet powerful description of how energy is distributed in frequency, enabling efficient signal processing and system design.