Digital filters: Design of FIR filters احسان احمد عرساڻي Lecture 23-24
Introduction to FIR filters These have linear phase No feedback Output is function of the present and past inputs only These are also called ‘all-zero’ and ‘non-recursive’ filters These do not have any poles
Applications Where: highly linear phase response is required Need to avoid complicated design
FIR Filter Design Methods Windows Frequency-sampling
FIR Filter Design: Windows Method Start from the desired frequency response Hd(ω) Determine the unit (sample) pulse reponse hd(n)=F-1{Hd(ω)} hd(n) is generally infinite in length Truncate hd(n) to a finite length M
Truncating hd(n) Take only M terms N=0 to N=M-1 Remove all others
Truncating hd(n) Take only M terms Remove all others N=0 to N=M-1 Remove all others Multiplying hd(n) with a rectangular window
Determine H(ω) Take Fourier transform of h(n) Therefore, compute: Hd(ω) and W(ω) Hd(ω) depends on the required response hd(n)
Computing W(ω) W(ω)=F{w(n)} w(n) is a rectangular pulse
Example A low-pass linear phase FIR filter with the frequency response Hd(ω) is required Hd(n) happens to be non-causal having infinite duration
The impulse response hd(n)
Windowing the hd(n)
The truncated hd(n)
Example A low-pass linear phase FIR filter with the frequency response Hd(ω) is required
Frequency of oscilation increases with M Magnitude of oscillation doesn’t increase or decrease with M Oscillations occur due to the Gibbs phenonmenon caused by the multipli- cation of the rectangular window with Hd(ω)
Frequency of oscilation increases with M Magnitude of oscillation doesn’t increase or decrease with M Oscillations occur due to the Gibbs phenonmenon caused by the multipli-cation of the rectangular window with Hd(ω)
Other windows
Other windows
Spectrum of Kaiser window (Cycles per sample)
Spectrum of Hanning window
Spectrum of Hamming Window (Cycles per sample)
Spectrum of Blackman Window (Cycles per sample)
Spectrum of Tukey Window (Cycles per sample)
Windows’ characteristics
The FIR filter’s response with Rectangular window M=61
FIR filter’s response with Hamming window
FIR filter’s response with Blackman window
FIR filter’s response with Kaiser window M=61
Using the FIR filter
Blackman’s filter output