Smoothing and filtering
Three things with similar names. Which one you want, and what each actually changes.
Smoothing a channel
The ∿ button on a channel row applies a centred moving average over a window you choose: Off, 50 ms, 100 ms, 150 ms, 250 ms, 500 ms, 1 s or 2 s. It takes the fuzz out of a noisy sensor without changing anything underneath: the raw data is kept and comes back with Off.
It applies to the graphs, the cursor readouts, the gauges, the statistics and the histograms, so a table built on a smoothed channel bins the smoothed values. It is saved with a layout, so a channel you always want smoothed stays that way.
The two kinds of filtering
- A histogram filter decides which samples are counted at all: throttle above 50, coolant above 170. Samples that fail are not in the table.
- Filtering in the estimated-acceleration settings is the width of the window used to differentiate speed. Wider is smoother and slower to react.
Smoothing changes how a value reads. A histogram filter changes what is counted. Acceleration filtering changes how an estimate is derived.
This page is part of the manual built into BigData for Windows, where the same topics are available offline with F1.