How do I low-pass filter a measurement signal?

Add the filter function fft_filter(rsFreqCutOffBelow, rsFreqCutOffAbove) to the curve. UniPlot transforms the signal with an FFT, removes all frequencies outside the range you give, and transforms it back. For a low-pass filter, set the lower limit to 0 and the upper limit to the cut-off frequency.

Signal with 50 Hz hum and noise, and the same signal after a 10 Hz low-pass filter

Signal with 50 Hz hum (gray) and the result of fft_filter(0, 10) (blue): only the components below 10 Hz remain.

How it works

  1. Make sure the x values of the curve are a time in seconds. The frequencies are given in Hz.

  2. Right-click the curve, choose Edit Filter Functions…, select fft_filter and click =>.

  3. Click Edit and enter the two limits. Everything below the first and above the second frequency is removed.

  4. Click OK.

In a script:

XYSetFilterFunctions(hData, "fft_filter(0, 10)");   // low pass 10 Hz

High-pass and band-pass

The same filter keeps any frequency band:

  • Low pass, keep 0 to 10 Hz: fft_filter(0, 10)

  • High pass, remove everything below 5 Hz: fft_filter(5, 1e10)

  • Band pass, keep 20 to 200 Hz: fft_filter(20, 200)

Why it helps

  • One filter covers low-pass, high-pass and band-pass.

  • The cut-off is given in Hz, so the setting is easy to check against the physics of the measurement.

  • The filter is stored in the dataset and recalculated for every new data file.

Good to know

  • The filter cuts the spectrum sharply at the limits. To find out which frequencies to remove, look at the spectrum first with fft_spectrum.

  • If the time steps are not constant, UniPlot interpolates the signal to constant steps before the FFT. The result has the same number of points, evenly spaced.

  • A high-pass filter also removes the mean value (0 Hz), so the result oscillates around zero.