Processing

Filtering

Reduce noise and restrict a curve to a potential window, time window, or CV cycle. The raw data is kept.

A filter should remove random fluctuation without broadening peaks or shifting their position.

Studio offers five filters: Savitzky-Golay for peak-preserving smoothing, moving average, median for isolated spikes, an FIR low-pass, and a notch for 50 or 60 Hz line pickup. All are zero-phase, so they do not shift peaks.

When to use it

Cleaning noisy DPV or SWV scans before peak finding.
Rejecting line-frequency pickup without distorting scan shape.
Smoothing a fast CV for presentation. The raw data stays in the file.
Avoid filtering when you plan to report peak widths; any filter broadens peaks slightly.

How it works

Savitzky-Golay fits a polynomial to a sliding window of points and replaces the centre point with the fitted value. The window width sets how aggressively noise is averaged out; the polynomial order sets how much peak shape is preserved.

The FIR (finite impulse response) filter convolves the data with a pre-designed kernel. Studio includes notch kernels for 50 Hz and 60 Hz and a configurable low-pass.

Parameters

Filter type

Typical: Savitzky-Golay

Savitzky-Golay, moving average, median, FIR low-pass, or notch. Savitzky-Golay is the default for voltammograms; median removes isolated spikes; notch removes mains pickup; FIR low-pass sets a sharp cut-off.

Window (SG)

Typical: 11 points

Number of points per fit. Must be odd. Larger windows smooth more but broaden peaks. Start near 5 % of your peak width in points.

Polynomial order (SG)

Typical: 3

2 or 3 for most scans. Order 3 preserves peak shape better; order 2 is more aggressive.

FIR kernel

Notch (50 / 60 Hz) or low-pass. Cutoff is user-configurable for the low-pass.

Cutoff (FIR low-pass)

Stop-band frequency in Hz. Choose above your signal bandwidth but below the noise you want to reject. FIR low-pass and notch both need a uniform sampling rate.

Range and cycle

Restrict the curve to a potential or time window, or a single CV cycle, before analysis. For EIS the range selects the frequencies the fit uses.

Example

Smooth a noisy DPV trace before peak finding

Load the noisy DPV scan. Peaks are visible but ragged.
Right-click the trace, choose Filter → Customize, and pick Savitzky-Golay. Window = 11, order = 3.
Overlay the filtered trace on the raw to confirm the peak is not softened.
Run peak finding on the filtered trace.
Export both the raw and the filtered trace.

Result. A cleaner trace where peak positions and widths are unchanged but baseline noise is cut by roughly a factor of two.

In Sensitify Studio

Right-click a curve and pick a Filter preset (Off, Light, Medium, Strong), or Customize to open the Data Processing window with its spectrum view. While the window is open the filtered trace is drawn over the raw one. The raw data stays in the file; setting the filter to Off restores it.

Step-by-step instructions: Data processing