Digital Signal Processing Reference
In-Depth Information
Oszilloscope
Noise
LP 2 00 Hz
FFT
BlockAver ag e
0,20
0,15
0,10
0,05
0,00
0,20
0,15
0,10
0,05
0,00
0
25
50
75
100
125
150
175
200
225
250
275
300
325
350
375
Hz
Illustration 121:
Noise as a test signal
The longer a noise signal lasts the less it is random with regard to the amplitudes of the frequencies
contained in the noise. This is proved by this measurement method.
Instead of a long noise signal the mean of the results of many noise signals of 1s in length is calculated in
the Illustration (top). At the bottom you see the spectrum of a lowpass filtered noise signal of 1s duration.
The noise is generated here by means of an algorithm (computational method) and strictly speaking no
longer corresponds to the ideal case. This is shown here because even after extremely lengthy block
averaging the filter curve is not “smooth”. In some places the peaks continue to exist.
