SIRIUS
®
TECHNICAL REFERENCE MANUAL
6.1.3.2. Aliasing
In signal processing, aliasing is an effect that causes different signals to become indistinguishable (or
aliases of one another) when sampled.
When signals are sampled , the sampling frequency must be at least twice as high as the maximum
frequency of the signal to avoid errors: this is called Nyquist theorem (Aliasing effect).
The illustration below shows an example of aliasing. The red sine wave is the original signal that we want
to measure. But our sampling frequency is too low (the sampling points are shown as black lollipops).
When we now reconstruct the sine wave by interpolating the sampled points, we get the blue signal,
which is also a sine-wave, but has the wrong frequency!
Aliasing effect
Real world measurement signals are hardly ever pure sine waves and can thus have many components
(harmonics) above the Nyquist frequency. These harmonics are erroneously aliased back to the
baseband and thus added to parts of the accurately sampled signal which produces a distorted
measurement signal. Filtering can be used to block frequencies above the Nyquist bandwidth (which is
half of the sample rate), to get correct measurement results.
Hint
To get a better understanding of the aliasing effect, we recommend to take the DEWESoft® PRO
training course: Spectral analysis using the FFT:
https://training.dewesoft.com/online/course/fft-spectral-analysis
Our professional online training courses are available free of charge to registered users of our
homepage.
Important
To avoid aliasing, we must cut all frequencies that are higher than half of the sample rate. But on
the other hand, we do not want to damp the amplitude of the signal below the Nyquist
frequency. To meet both goals, we need a filter (in this context called: anti-aliasing filter) with
sharp damping at half of the sampling rate (which requires a high order filter).
SIRIUS
®
V20-1 269 / 336
To Next Page
Loading...
Need help?
General
