Prev: MEX Setup for Mac OS X 10.6
Next: execute a matlab script from a web page
From: Chris on 11 Feb 2010 16:51
"John " <jwu0823(a)hotmail.com> wrote in message <gli8kk$ltk$1(a)fred.mathworks.com>...
> I got it! The biggest mistake I made was that I made f fixed and mixed up with the sampling frequency.
> Thanks for all the help, especially to Greg.

I'm trying to get the same thing working and am having problems too.

clear all

Fs = 100;
dt = 1/Fs;
t0 = pi/2;

t = 0:dt:3*pi;
l = max(size(t));
df = 1/dt;%dt*2*pi;
dw = 2*pi*df;
f = (-1*length(t)/2:1:length(t)/2-1)*df;
f = fftshift(f);
w = 2*pi*f;
w = fftshift(w);

x = sin(t);
x2 = sin(t-t0);

X = fft(x);
X = X.*exp(-i*2*pi*t0*f/length(t));
y = ifft(X);

plot(t,x)
hold on
plot(t,x2,'r--')
plot(t,y,'g')

is my code and what should be happening is that the output y should matchup with the x2 however, there is some rectification as well as attenuation going on. any ideas?
From: Matt J on 16 Feb 2010 23:25
"Chris " <news(a)comcast.com> wrote in message <hl1u46$cib$1(a)fred.mathworks.com>...
> "John " <jwu0823(a)hotmail.com> wrote in message <gli8kk$ltk$1(a)fred.mathworks.com>...
> > I got it! The biggest mistake I made was that I made f fixed and mixed up with the sampling frequency.
> > Thanks for all the help, especially to Greg.
>
> I'm trying to get the same thing working and am having problems too.
====================

I hope this is homework, rather than an actual application. Time shifting using FFTs is super-inefficient.
From: Greg Heath on 16 Feb 2010 19:31
On Feb 11, 4:51 pm, "Chris " <n...(a)comcast.com> wrote:
> "John " <jwu0...(a)hotmail.com> wrote in message <gli8kk$lt...(a)fred.mathworks.com>...
> > I got it! The biggest mistake I made was that I
> made f fixed and mixed up with the sampling frequency.
> > Thanks for all the help, especially toGreg.
>
> I'm trying to get the same thing working and am having
> problems too.
>
> clear all
>
> Fs = 100;
> dt = 1/Fs;
> t0 = pi/2;
>
> t = 0:dt:3*pi;

t = dt*(0:N-1) = 0:dt:T-dt % T = N*dt

> l = max(size(t));
> df = 1/dt;%dt*2*pi;

No. See above that Fs = 1/dt

df = 1/T = 1/(N*dt)

> dw = 2*pi*df;
> f = (-1*length(t)/2:1:length(t)/2-1)*df;
> f = fftshift(f);

No. fftshift is only used on transforms

> w = 2*pi*f;
> w = fftshift(w);

No. fftshift is only used on transforms

> x = sin(t);
> x2 = sin(t-t0);
>
> X = fft(x);
> X = X.*exp(-i*2*pi*t0*f/length(t));
> y = ifft(X);
>
> plot(t,x)
> hold on
> plot(t,x2,'r--')
> plot(t,y,'g')
>
> is my code and what should be happening is that the output y should matchup with the x2 however, there is some rectification as well as attenuation going on. any ideas?

See my post.

Hope this helps.

Greg
From: Chris on 17 Feb 2010 10:25
"Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message <hlfr3a$97g$1(a)fred.mathworks.com>...
> "Chris " <news(a)comcast.com> wrote in message <hl1u46$cib$1(a)fred.mathworks.com>...
> > "John " <jwu0823(a)hotmail.com> wrote in message <gli8kk$ltk$1(a)fred.mathworks.com>...
> > > I got it! The biggest mistake I made was that I made f fixed and mixed up with the sampling frequency.
> > > Thanks for all the help, especially to Greg.
> >
> > I'm trying to get the same thing working and am having problems too.
> ====================
>
> I hope this is homework, rather than an actual application. Time shifting using FFTs is super-inefficient.

How would you suggest to time shift then.
From: Matt J on 17 Feb 2010 11:04
"Chris " <news(a)comcast.com> wrote in message <hlh1on$mq$1(a)fred.mathworks.com>...
> "Matt J " <mattjacREMOVE(a)THISieee.spam> wrote in message <hlfr3a$97g$1(a)fred.mathworks.com>...
> > "Chris " <news(a)comcast.com> wrote in message <hl1u46$cib$1(a)fred.mathworks.com>...

>
> How would you suggest to time shift then.

interp1(t,x,t+t0)

This is an O(N) computation using only real arithmetic, whereas with FFTs it's
an O(N*logN) computation using complex arithmetic.
 |  Next  |  Last
Pages: 1 2
Prev: MEX Setup for Mac OS X 10.6
Next: execute a matlab script from a web page