conv from conv function and ifft

I am trying to find the conv for a specific impulse and signal. Now when I try to use the conv function I ge the answer I need. But when I try to find fft for the impulse and then fft for the signal then .* both fft then ifft the answer I get a different answer. I tried many times any idea why?

Thanks in advance.

Answer 1

[Matt J]

Most likely a zero-padding problem.

https://www.mathworks.com/help/signal/ug/linear-and-circular-convolution.html

[Omar Abulayla]

I tried to convert the cconv to linear conv so I did

l1=length(x); 
l2=length(h); 
N=max(l1,l2); 
xpad=[x zeros(1,N-l1)]; 
hpad=[h zeros (1,N-l2)]; 
ccirc = ifft(fft(xpad).*fft(hpad)); 

I got the exact results from the fft the 1st time, although I am pretty sure its incorrect.

[Matt J]

N=(l1+l2-1);

[Omar Abulayla]

Error using horzcat Dimensions of arrays being concatenated are not consistent.

[Matt J]

The length has to be l1+l2-1 because that is the expected length of the output with conv().

[Omar Abulayla]

I got this error ‘’Error using horzcat Dimensions of arrays being concatenated are not consistent’’

[Matt J]

The error has nothing to do with the code you've shown, or my proposed modification:

 x=rand(1,3); h=rand(1,5);

 l1=length(x); l2=length(h); 
 N=l1+l2-1;
 xpad=[x zeros(1,N-l1)]; 
 hpad=[h zeros(1,N-l2)]; 
 
 ccirc = ifft(fft(xpad).*fft(hpad))

ccirc = 1×7

0.1116 0.2622 0.4966 0.5405 0.5278 0.3675 0.1407

●

[Image Analyst]

This works fine, so what did you do?

fprintf('Beginning to run %s.m ...\n', mfilename);
% Make random signal.
signalLength = 32;
x = rand(1, signalLength);
% Make impulse function:
h = zeros(1, signalLength);
h(signalLength/2) = 1;
% Pad the signal and kernel.
l1=length(x)
l2=length(h)
N = max(l1,l2)
xpad=[x, zeros(1,N-l1)]; 
hpad=[h, zeros(1,N-l2)]; 
ccirc = ifft(fft(xpad).*fft(hpad)); 
fprintf('Done running %s.m.\n', mfilename);

[Matt J]

Looks fine. No error messages.

[Matt J]

Please post your code in code wells, so that it can be more easily read, copied, and run.

Aside from that, why are we talking about your early version that doesn't incorporate my modification? I've already told you that that version had insufficient zero-padding for that to work.