What is a Blackman -61dB filter used for?

[jcannon]

Hi everyone,

I am seekling some assistance with determining if a Blackam -61dB bandpass filter (10-500Hz) is suitable for determining the amplitude of both voluntary and evoked EMG signals...

Typically, I use 2nd or 4th order butterworth filters for processing my emg signals (usually 20-450Hz, but sometimes 10-1000Hz depending on the type of signal)... I have read online that the Blackman filter boundaries are not very square (lots of roll off) and they are not good for amplitude measurements (ok for frequency though)... The higher order butter filters provide sharper cut-offs.

Can some please comment on this?

Thanks.

Jack

[johnsold]

Jack,

The term Blackman -61 dB filter is rather meaningless. First the filter type is separate from any attenaution specification. Second the attenuation cannot be specified by a single number. You need to specify the frequency at which the attenuation is required to occur.

Butterworth filters have very smooth attenaution versus frequency curves but relatively slow rolloff as a consequence of the smoothness. Chebshev filters have steeper rolloffs but have ripple or varaitions in the attenuation in the passband - they are lss smooth than Butterworth. Elliptical filters have much steeper rolloffs at the expense of ripples in both the pass band and stop band.

The comparison of the frequency repsonses of filters should always be done with the same order for each filter type. For the selection of a filter type given a particular specification a determination of the order required to meet the specified attenautation may result in the choice of a high order Butterworth filter or a lower order elliptical filter.

I have four filter books on my desk and none of them mention Blackman. Blackman windows are used in spectral analysis to reduce spectral leakage from the finite nature of the measurement of real signals.

Most EMG signals do not have a lot of energy outside the desired bandwidth so the filter selection may not be too critical. One consideration is that the use of a different filter from what others doing similar work have used will make it very difficult to accurately compare results.  Unfortunately, many of the standard references in the field seem to have been developed by people who knew nothing about the frequency characteristics of the signals they were measuring or the equipment they were using.

Lynn

[jcannon]

Thanks, Lyn...