Envelope, phase, filters, and decimation
The methods of this chapter use Primary measured signal. Empty the possible coupled channel. The axis of the results is in seconds if the sampling rate is known. With measured time, the first value retains the initial coordinate converted in seconds. With a declared frequency and empty coordinate, the axis starts at zero.
analytic.envelope — analytic envelope
Select analytic.envelope and run Analyze. The magnitude of the Hilbert analytic signal estimates the amplitude envelope. This is useful for a modulated carrier that is well separated in frequency. amplitude uses the signal unit. The horizontal coordinate is time_seconds, or a zero-based sample_index when the rate is unknown, with unit 1. In the modulated exercise, the envelope varies from about 0.6 to 1.4 V. Consider beating and edge artifacts when components overlap or discontinuities are present.
analytic.phase — analytic phase
With analytic.phase, phase_rad is the argument of the analytic signal in radians. The phase is unwrapped and can exceed 2π. Use the time_seconds or sample_index axis. For a 32 Hz tone sampled at 256 Hz, phase increases by about 0.7854 rad per sample. The initial offset depends on the waveform and convention. Phase is hard to interpret when amplitude approaches zero or several components compete.
analytic.instantaneous_frequency — instantaneous frequency
Select analytic.instantaneous_frequency with measured time or a positive declared rate. The method takes the time derivative of unwrapped phase and divides it by 2π. frequency_hz is in Hz and time_seconds in seconds. The test tone stays near 32 Hz. Noise, edges, and amplitude minima can amplify apparent variation. Interpret the result as a local frequency where a dominant component can be identified.
Butterworth low-pass cutoff [Hz] and Butterworth order
Select filter.causal or filter.zero_phase. Set Butterworth low-pass cutoff [Hz], initially 10, strictly between zero and Nyquist. At 256 Hz you have to choose less than 128 Hz. The cutoff refers to the single-pass design. Butterworth order accepts integers from 1 to 6, with 3 default. Higher orders make the transition sharper but can accentuate transient and numerical sensitivity. The respective Reset buttons restore 10 and 3.
filter.causal — causal low-pass filter
Set rate, cut and order, then Analyze. The IIR filter uses the current sample and state of past samples, with zero initial state. Use this method to study a causal response. It can delay and deform the phase depending on the frequency. Check the startup transient. The output retains one row per sample and time_seconds columns in s and value_real in the signal unit.
filter.zero_phase — zero-phase low-pass filter
Set the sample rate, cutoff, and order, then select Analyze. Filtering forward and backward compensates for phase shift away from the edges. This method uses future samples and is intended for completed acquisitions. Its magnitude response is the square of the single-pass response, increasing attenuation. Edge artifacts remain possible. The series must be long enough for reflection at the boundaries. The backend typically uses more than three times the order. The output columns are the same as for the causal filter.
multirate.decimate — decimation with anti-aliasing
Enter a valid sample rate and set Decimation factor to an integer from 2 to 64. The default is 2 and Reset restores it. Analyze applies an anti-aliasing filter before reducing the sample count. The output rate is the input rate divided by the decimation factor. With 1,024 rows at 256 Hz and a factor of 2, you get 512 rows at 128 Hz. The new Nyquist frequency is 64 Hz, so the exercise’s 80 Hz component should be strongly attenuated.
time in seconds starts from the available source and advances with the new step. value_real retains the input unit. The summary includes up_factor, down_factor and output_sample_rate. Unlike Display source, the decimation table contains a processed signal, with fewer rows, a different sample rate, and a narrower useful band.