In this experiment digital filter is designed using Frequency sampling Method(FSM).
If the filter is a linear phase FIR filter then Φ= -(N-1)ω/2 where N is the length of h[n].
The given data should be : type of filter(LPF or HPF), Order of the filter and the digital cutoff frequency (ωc).If we compare it with other methods then the difference is that we find H[k] by sampling using DTFT and then we obtain transfer function h[n] by IDFT method.
In case of non linear phase filter value of Φ=0.Its simpler to implement if compared to other methods.
If the filter is a linear phase FIR filter then Φ= -(N-1)ω/2 where N is the length of h[n].
The given data should be : type of filter(LPF or HPF), Order of the filter and the digital cutoff frequency (ωc).If we compare it with other methods then the difference is that we find H[k] by sampling using DTFT and then we obtain transfer function h[n] by IDFT method.
In case of non linear phase filter value of Φ=0.Its simpler to implement if compared to other methods.
This is the simplest method to design FIR filter.
ReplyDeleteYes, FSM is simpler because many samples in the input impulse function are zero, which reduces the number of calculatuons
ReplyDeleteYes it is a better method
ReplyDeleteClear concept
ReplyDeletewhy do we prefer FSM instead of windowing method?
ReplyDeleteThe window method is based on calculating the impulse response of an ideal digital filter, which is IIR, and on using a window to obtain a finite impulse response. One of the most interesting properties you can get is linear phase.
DeleteFrequency sampling methods only guarantee good approximation to the desired response in some frequency values. The approximation in other values can be very bad.
Frequency sampling algorithm requires minimum information about the filter in the input sufficient in design of the filter.
Windowing method is preferred more because the characteristics of the filter is decided by the window function.
DeleteFor optimum output,windowing method is better.But in terms of no. of calculations FSM is preferred because it is carried out for discrete signals whereas windowing method requires continuous time domain calculations.
DeleteHowever, if undersampled, output may be time aliased!
ReplyDeletewell explained
ReplyDeleteFSM involves discrete computation compared to continuous used in windowing method. Hence, the faster implementation. However,as frequency sampling is involved, the sampling frequency must be selected such that the samples contain useful data.
ReplyDeleteHowever, if undersampled, output can be time-aliased!
ReplyDelete