

Journal of Electromyography and Kinesiology, February 1999 9(1):III-IV Wireless Engineer (also called Experimental Wireless and the Wireless Engineer), vol. Filtering the surface EMG signal: Movement artifact and baseline noise contamination. LabChart Arithmetic Lowpass channel calculation settings.

at 2kS/s sampling rate, the high-pass transition width is 15 Hz, which corresponds approximately to a 7th order Butterworth with a half-power cut-off of 21.6 Hz. frequency response) changes for higher sampling rates. However, at 100 kS/s and above, the filter becomes less sharp than the 4th order Butterworth. Please note that with the values of fc = 2.865 Hz and df = 3.2 Hz, the Arithmetic expression above is valid for sample rates up to 40 kS/s. In LabChart, with an Arithmetic channel calculation (either on the EMG source channel or in a new channel) you would have something like this then: LabChart Arithmetic channel calculation settings Frequency responses for both the 4th order Butterworth and the FIR filter for the case where the desired Butterworth filter cut-off frequency (fcB) = 2.5 Hz. However, a FIR low-pass filter that approximates the 4th order Butterworth low-pass frequency response in the range of greatest importance (0 dB to -12dB), can be created by setting the FIR cut-off frequency to 1.146 times the desired Bufferworth cut-off frequency (fcB) and setting a “User defined Transition width” which is 1.28 times fcB.įor a 2.5 Hz Butterworth you would use a cut-off of 2.865 Hz, and a Transition width of 3.2 Hz. LabChart currently does not support zero-phase-lag-Butterworth filters. Approximating a 4th order Butterworth Filter This would correspond approximately to a 4th order Butterworth high-pass filter at 22.4 Hz. for a sampling rate of 4kS/s, the high-pass transition width is 30 Hz. This would correspond approximately to a 24th order Butterworth low-pass filter with a half-power cut-off frequency of 490 Hz.įor the high-pass component of the 20-500 Hz band-pass filter in the “Auto adjust” mode, the transition width depends on the sampling rate. If you use the “Auto adjust” setting for the “Transition Width”, the low-pass filter will have a transition width of 0.2*500 Hz = 100 Hz. The LabChart band-pass filter is comprised of a high-pass FIR filter with a half-amplitude frequency of 20 Hz and a low-pass FIR filter with half-amplitude frequency of 500 Hz. For the LabChart digital filters, this is the frequency range for which the output amplitude is between 1% and 99% of the input amplitude. As a result, the sharpness of FIR filters is described by specifying the “Transition Width”. Instead the slope continues to become steeper until the stop-band ripple level is reached. Īlso, unlike IIR filters, FIR filters do not asymptotically approach a constant attenuation slope on a log-log plot. The half-amplitude point is used because the transition region of FIR filters is anti-symmetric about this point. For FIR filters, the cut-off frequency used for the design is conventionally the half-amplitude frequency rather than the half-power frequency used for IIR filters. The design parameters specified for FIR filters are different from those used for Infinite Impulse Response (IIR) filters such as Butterworth filters. Digital FIR phase filters are recommended for processing EMG signal amplitude. not a function of frequency, and so they can also be used online in real-time. Unlike Butterworth filters, these filters are “linear-phase”, which means the delay introduced by the filter is a constant (actually 0 in LabChart), i.e.

These FIR filters are designed using the “Window Method” with a Kaiser window with beta = 6, which results in pass and stop band ripple of less than 0.5%. The standard digital filters supplied by the LabChart Digital Filter Channel Calculation are zero-phase-lag Finite Impulse Response (FIR) filters. A Butterworth filter is recommended for general sEMG signal filtering use - it is frequency-based and has a faster setting time in response to signal transients, and the effect of filtering can be easily understood and predicted. The simplest and most direct means of increasing the fidelity of the sEMG signal is to filter the maximum amount of noise while retaining as much of the desired EMG signal frequency spectrum as possible. The surface electromyographic (sEMG) signal that originates in the muscle is contaminated by various noise signals or artifacts.
