Filter Acceleration as Impacts Only

If this option is selected, a real-time filter is applied so that the amplitude of impacts on the sensor are displayed and stored instead of the raw data from the sensor.

The algorithm used for computing the filtered values is as follows:

1. The data are high-pass filtered with a 10 Hz finite-impulse-response (FIR) filter, using a Blackman window and a filter kernel size of 127.  The latency introduced by this filter is approximately 0.86 seconds when operating at the low data rate (2000 Hz EMG / 148 Hz ACC) and approximately 0.43 seconds when operating at the high data rate (4000 Hz EMG / 296 Hz ACC).  This filter removes the steady-state bias component of the acceleration.  Note that impacts that happen at a rate of 10 Hz or lower will be attenuated by this filter.
2. The data are rectified.
3. The data are low-pass filtered with a 5 Hz finite-impulse-response (FIR) filter, using a Blackman window and a filter kernel size of 127.  The latency introduced by this filter is approximately 0.86 seconds when operating at the low data rate (2000 Hz EMG / 148 Hz ACC) and approximately 0.43 seconds when operating at the high data rate (4000 Hz EMG / 296 Hz ACC).  This filter smoothes the impacts so that they can be easily thesholded in post-processing.

Each output channel is assigned units corresponding to rectified acceleration data (please note that this data is significantly band-limited by processing).  Impacts are filtered independently and simultaneously on each axis of the accelerometer.  For easy comparison when using this option, the overall latency introduced by the impact and inclination real-time filters is identical.  The latency may be corrected for with the TIME SHIFT calculation in EMGworks analysis.

This option is available to all sensors that contain acceleration data channels [Standard Trigno, Snap Lead, Spring Contact, EKG]