How does Signal Truncation affect FFT results, particularly in terms of Spectral Leakage?
In practical signal processing, we never observe signals over infinite time. Instead, we analyze a finite-length (truncated) segment of the signal. This truncation has a direct and important effect on FFT results. It introduces spectral leakage.
What Is Signal Truncation?
Truncation = windowing in time domain
where
- x(t) original (infinite) signal
- w(t) window (often rectangular)
In FFT, We multiply the signal by a finite window, typically a rectangular window
Why Truncation Causes Spectral Leakage?
Time–Frequency Duality (convolution theorem)
Multiplication in time ↔ convolution in frequency
where:
- X(f) true spectrum
- W(f) window spectrum
Rectangular Window Effect
Relationship of xw (t) and Xw (f)
※ The FFT results shown above corresponds to the Continuous / Discrete-Time Fourier Transform(DTFT), while the FFT represents its sampled version at discrete frequency bins, as illustrated in the example below.
How to Reduce Spectral Leakage?
Coherent Sampling
- Ensure integer number of cycles
- Align frequencies with FFT bins
Windowing
Apply smooth windows
- Hann window
- Hamming window
- ...
Effect
- Reduces side lobes
- But increases main lobe width (smear)
Longer Record Length
Longer signal
- Better frequency resolution
- Less leakage impact
Practical Example
Coherent sampling: sine wave truncated to an integer number of periods
Leakage-free FFT occurs: DTFT (continuous) vs FFT (sampled)
Real-world sampling: sine wave truncated to an non-integer number of periods
Leakage FFT occurs: DTFT (continuous) vs FFT (sampled)
Key Insight
Spectral leakage is not a numerical error—it is a mathematical consequence of data truncation. A finite observation (record length) window inevitably distorts the spectrum. In the DTFT, which represents the continuous spectrum, truncation spreads spectral energy in a sinc-shaped form. In FFT analysis, leakage becomes visible when the signal frequency is not coherently aligned with the FFT bins (non-coherent sampling). In practice, spectral leakage usually refers to this leakage observed in the FFT spectrum.
Conclusion
Signal truncation causes spectral leakage because it introduces discontinuities, spreading energy across frequency bins.
In summary
- Truncation = windowing
- Windowing → convolution in frequency
- Convolution → sinc-shaped energy spreading
- Non-coherent sampling → spectral leakage in FFT
Suggested Further Reading
##You may also find these topics helpful:
How does Signal Truncation affect FFT results, particularly in terms of Spectral Leakage?
In practical signal processing, we never observe signals over infinite time. Instead, we analyze a finite-length (truncated) segment of the signal. This truncation has a direct and important effect on FFT results. It introduces spectral leakage.
What Is Signal Truncation?
Truncation = windowing in time domain
where
In FFT, We multiply the signal by a finite window, typically a rectangular window
Why Truncation Causes Spectral Leakage?
Time–Frequency Duality (convolution theorem)
Multiplication in time ↔ convolution in frequency
where:
Rectangular Window Effect
Relationship of xw (t) and Xw (f)
※ The FFT results shown above corresponds to the Continuous / Discrete-Time Fourier Transform(DTFT), while the FFT represents its sampled version at discrete frequency bins, as illustrated in the example below.
How to Reduce Spectral Leakage?
Coherent Sampling
Windowing
Apply smooth windows
Effect
Longer Record Length
Longer signal
Practical Example
Coherent sampling: sine wave truncated to an integer number of periods
Leakage-free FFT occurs: DTFT (continuous) vs FFT (sampled)
Real-world sampling: sine wave truncated to an non-integer number of periods
Leakage FFT occurs: DTFT (continuous) vs FFT (sampled)
Key Insight
Spectral leakage is not a numerical error—it is a mathematical consequence of data truncation. A finite observation (record length) window inevitably distorts the spectrum. In the DTFT, which represents the continuous spectrum, truncation spreads spectral energy in a sinc-shaped form. In FFT analysis, leakage becomes visible when the signal frequency is not coherently aligned with the FFT bins (non-coherent sampling). In practice, spectral leakage usually refers to this leakage observed in the FFT spectrum.
Conclusion
Signal truncation causes spectral leakage because it introduces discontinuities, spreading energy across frequency bins.
In summary
Suggested Further Reading
##You may also find these topics helpful: