Frequency/Magnitude response of lowpass FIR filter

[Naveed Ahmed]

Hi,

I have written a test program in matlab for lowpass fir, when i m plotting its magnitude response after fft, it seems to be a high pass filter. Why its doing so??
My code is as follow:

N=50; %Filter order
Length=N+1; %length
n0=N/2; %delay parameter
n=0:1:N;
fc_l=1/8; %Cut off frequency for the low pass filter
Wc_l=2*pi*fc_l;
%===================LOWPASS FIR FILTER=====
h_lp=sin(Wc_l*(n-n0)) ./(pi*(n-n0)); %Impulse response of Low Pass FIR filter For n-n0!=0
h_lp(n0+1)=(Wc_l/pi) %Imp response of Low Pass FIR filter For n=n0 or n-n=0
F=fft(h_lp,51);
X=real(F);
Y=imag(F);
[th,Mag]=cart2pol(X,Y);
plot(0:50,Mag)

Thnks,

ur comments plz

naveed

 

[chahid]

Hi,

if you do this
f=(0:50)/50
plot(f,Mag)

your plot will be normalized and you see the relevant part which is 0 to 0.5
the part higher than 0.5 is the negative spectrum (0.5 is the nyquist frequency)
your filter is correct.
best regards

---------- Post added at 09:21 ---------- Previous post was at 08:53 ----------

Hi Naveed,
here is a slight modification that will show the spectrum of your filter

N=50; %Filter order
Length=N+1; %length
n0=N/2; %delay parameter
n=0:1:N;
fc_l=1/8; %Cut off frequency for the low pass filter
Wc_l=2*pi*fc_l;
%===================LOWPASS FIR FILTER=====
h_lp=sin(Wc_l*(n-n0)) ./(pi*(n-n0)); %Impulse response of Low Pass FIR filter For n-n0!=0
h_lp(n0+1)=(Wc_l/pi) %Imp response of Low Pass FIR filter For n=n0 or n-n=0

F=fft(h_lp,1024); %% 1024 higher frequency resolution
F=fftshift(F); % moving the zero-frequency component to the center
% X=real(F);
% Y=imag(F);
% [th,Mag]=cart2pol(X,Y);
% plot(0:1023,Mag)
f=(-512:511)/1024 % Normalised Frequency vector (-0.5 to +0.5)
plot(f,abs(F))

 

[Naveed Ahmed]

Thanks chahid,

Its showing now, there is another method aswell, we can use the builtin freqz command... freqz(h_lp)
But it shows us in normalized frequency mode, we need to convert that. Whats the conversion relation between normalized freq to hz??

Thx

 

[chahid]

Hi Ahmed,


In normalized spectrum means sampling frequency = 1 you have just to multiply by
the sampling frequency to get it in hz. or use directly
freqz(h_lp,1,1024,Fs) Fs is sampling frequency if you put Fs=1000 hz you get
the cut frequency = fc_l * Fs