An comb filter, or sinc filter, is typically used in highly oversampled situations because it has a monotonically decreasing passband ("droop") which is much sloppier than an equiripple FIR design. The advantage of a comb filter is that it can implement an extremely high dynamic range (passband-gain / stopband-gain) using only relatively low-precision integer coefficients. The disadvantage is that the stopbands are relatively narrow, and the passband has the droop already mentioned. Comb filters are various cascades of the extremely simple impulse response [ 1 1 1 .... 1 ], where the length N of the impulse response is usually the decimation/interpolation ratio. Your data is scarcely oversampled, and as such I would not use a comb filter. Or perhaps when you say "comb filter" you really mean an FIR filter, whereas I think of the specific filter I've been describing. The best filter for you would be one which has a stopband at 60 Hz and a passband over your data frequencies. You can use "sptool", which I believe is part of the Matlab signal processing toolbox, to design such a filter. Sincerely, Glen Ragan |