QuantAsylum QA401, Руководство пользователя

Страница: 30 / 70

QuantAsylum QA401 User ’s Manual
Page
30
Windowing Functions
The FFT process assumes the captured data repeats infinitely. If the data sent to the FFT function
contains, for example, 10.5 cycles of a waveform, then the extra half cycle will causes errors in the
resulting spectrum. To get around this, a windowing function is applied to the data. This function will
gently taper the amplitude of the collected data (prior to FFT) down towards 0 near the start and finish
of the sample buffer. By doing this, the extra half cycle that appears to immediately truncate will be
suppressed. The downside to this is that some other distortions will be introduced into the displayed
spectrum. But the introduced distortions are much less objectionable than the distortions that arise
from an abruptly terminated waveform.
Using the input display feature on the QA401, we can look at this in more detail. Using a small buffer
(512 samples) and low frequency waveforms, we can see what truncations can do to our data. For these
tests, we set the windowing function to Rect (which means no windowing is applied).
First, let’s look at a 10 Hz waveform at 16K points, 48Ksps. We c an see the input waveform below