FFT Measurements

Subtopics

These measurements can be made on FFT waveforms only.

To create an FFT waveform, use the FFT math function, which converts a time domain waveform into a frequency domain waveform. See FFT Operator.

The Infiniium application performs the FFT operation on the sample points that are in the waveform window. If the value is not a power of two number of points, the Infiniium application adds zeros equally to the beginning and end of the data until there are a power of two number of points. This is called the Effective Memory Depth.

The FFT resolution is defined as sampling rate / memory depth when using the Rectangular window (other windows have a Normalized Equivalent Noise Bandwidth factor applied).

You can increase the FFT resolution during data capture by manually setting the memory depth to a power of two. See Acquisition Tab.

The FFT frequency range, before scaling, is 0 Hz to 1/2 the sampling frequency. The sampling rate must be at least twice the highest frequency component in your waveform. Otherwise, the measurement results will exhibit aliasing and any measurements may be inaccurate.

Because peaks vary widely depending upon the Peak Level, you should set the Peak Values after you have a stable FFT for the best measurements. The Peak Level is specified when you add or edit an FFT measurement.

FFT Measurement Tips

  • For all FFT measurements:

    • When capturing waveform data, set the oscilloscope acquisition sampling mode to Real Time and set the sin(x)/x interpolating filter to off. See Acquisition Tab.

    • When capturing waveform data, set the oscilloscope vertical sensitivity of the source waveform so that the input waveform is near full screen but not clipped.

      Set the horizontal scale (or manually set the sampling rate and memory depth) so as many cycles of the waveform as possible are visible on the display. See Timebase Setup.

    • The FFT resolution is set by the oscilloscope sampling rate and memory depth used when capturing data. If both sampling rate and memory depth are set to automatic, the horizontal scale can affect both sampling rate and memory depth.

    • Set up the FFT peak definitions when adding or editing an FFT measurement.

  • For best frequency accuracy on peaks (for FFT frequency measurements):

    • Select the Hanning window when setting up the FFT function. See FFT Operator.

      You can turn off the source waveform to better see the FFT function.

    • The frequency accuracy is the sum of two terms:

      • The first term is the result of a limited number of frequency bins. Frequency bins are the N/2 frequency domain points that are output from the FFT of N time domain samples. Frequency bins are spaced equally in frequency, and the frequency step is the reciprocal of the time record length. The measurement is accurate to plus or minus half a bin.

      • The second term is related to the accuracy of the internal oscillator that generates the sample clock.

  • For the best vertical accuracy on amplitude measurements (for FFT magnitude measurements):

    • Select the Flattop window when setting up the FFT function. See FFT Operator.

      You can turn off the source waveform to better see the FFT function; however, keep in mind the FFT is performed only on the portion of the source waveform you see on the display.

  • When using FFTs, make sure you avoid waveform aliasing. Aliasing occurs when there are insufficient samples on each cycle of the input waveform to reconstruct the waveform. It occurs whenever the frequency of the input waveform is greater than the Nyquist frequency, which is the sample frequency divided by 2.

    When a waveform is aliased, it shows up in the FFT spectrum as a waveform of a lower frequency. Because the frequency span goes from 0 to the Nyquist frequency, the best way to prevent aliasing is to make sure that the frequency span is greater than all the frequencies present in the input waveform. Keep in mind that most periodic waveforms that are not sine waves have frequency components much higher than the fundamental frequency of the waveform. Those components may cause aliasing also.