From: Matt kloepfer on
Hello,

My team is working on a signal processing project. We are modeling our system in Simulink. We generate our own input data and pass this into the system using the From File Block where we input the .mat input signal.

The next step is to pass this input array through an FFT. We are using the FFT block provided in Simulink. Our problem is this: the array is FFT'd point by point, in other words, in real time. Instead, we would like to input the entire array and then FFT the entire array at once.

We tried using a buffer and that didn't work. How can we "hold" the inputs so that they are NOT computed in real time? Is this a setting within the FFT? Is this a problem with the input data? What about Simulation settings?

Thanks,

Matt
From: Navan Ruthramoorthy on
"Matt kloepfer" <kloepm(a)gmail.com> wrote in message <hpktib$hir$1(a)fred.mathworks.com>...
> Hello,
>
> My team is working on a signal processing project. We are modeling our system in Simulink. We generate our own input data and pass this into the system using the From File Block where we input the .mat input signal.
>
> The next step is to pass this input array through an FFT. We are using the FFT block provided in Simulink. Our problem is this: the array is FFT'd point by point, in other words, in real time. Instead, we would like to input the entire array and then FFT the entire array at once.
>
> We tried using a buffer and that didn't work. How can we "hold" the inputs so that they are NOT computed in real time? Is this a setting within the FFT? Is this a problem with the input data? What about Simulation settings?
>
> Thanks,
>
> Matt

Hi Matt,

Buffering is the usual way to create a frame of input data from scalars which then can be fed to FFT block. When you say "buffer didn't work" what was the problem you had?

Navan
From: Matt kloepfer on

>
> Hi Matt,
>
> Buffering is the usual way to create a frame of input data from scalars which then can be fed to FFT block. When you say "buffer didn't work" what was the problem you had?
>
> Navan



Hi Navan,

Thanks for the suggestion. We thought a buffer was the right way to go as well, but, our system is still not running like it should. The buffer is not holding the data like we want it to. The input is 32,768 points long. The first row is 1 to 32,768 and the second row is our data (we have tried the transpose of this as well). Then, we want to run the fft on all 32,768 data points at the same time.

We coded our system in a .m file. The goal is to make the .m file output match that of Simulink. Here is a link containing some photographs. "Output" is the picture of our Simulink output from scope 2, as seen in "Schematic." "Desired" is the output we obtained with our .m file. The .m file is confirmed correct, but, we can not make our Simulink output match this.

Is it the settings in the buffer? We have read the documentation for the buffer, but, are still left having difficulties.

Thanks,

Matt

The link is http://rpi.edu/~brankd
From: meijuan on
prada sneakers:
http://www.brandclothingstore.com/wholesale/Prada-Sneakers-548_p1.html
louis vuittion sneakers:
http://www.brandclothingstore.com/wholesale/Louis-Vuitton-Sneakers-549_p1.html
Apple Bottoms clothing:
http://www.brandclothingstore.com/wholesale/Women-Apple-Bottoms-Clothing-637_p1.html
prada clothing:
http://www.brandclothingstore.com/wholesale/Womens-Prada-Clothing-670_p1.html
coogi clothing:
http://www.brandclothingstore.com/wholesale/Coogi-Clothing-683_p1.html
chanel clothing:
http://www.brandclothingstore.com/wholesale/Chanel-Clothing-684_p1.html
gucci sneakers:
http://www.brandclothingstore.com/wholesale/Women-Gucci-Sneakers-606_p1.html
jordan sneakers:
http://www.brandclothingstore.com/wholesale/Women-Jordan-Sneakers-607_p1.html
jordan sneakers:
http://www.brandclothingstore.com/wholesale/Jordan-Sneakers-546_p1.html
new jordan 2010:
http://www.brandclothingstore.com/wholesale/New-Jordan-2010-1125_p1.html
From: Navan Ruthramoorthy on
"Matt kloepfer" <kloepm(a)gmail.com> wrote in message <hq8b9s$rii$1(a)fred.mathworks.com>...
>
> >
> > Hi Matt,
> >
> > Buffering is the usual way to create a frame of input data from scalars which then can be fed to FFT block. When you say "buffer didn't work" what was the problem you had?
> >
> > Navan
>
>
>
> Hi Navan,
>
> Thanks for the suggestion. We thought a buffer was the right way to go as well, but, our system is still not running like it should. The buffer is not holding the data like we want it to. The input is 32,768 points long. The first row is 1 to 32,768 and the second row is our data (we have tried the transpose of this as well). Then, we want to run the fft on all 32,768 data points at the same time.
>
> We coded our system in a .m file. The goal is to make the .m file output match that of Simulink. Here is a link containing some photographs. "Output" is the picture of our Simulink output from scope 2, as seen in "Schematic." "Desired" is the output we obtained with our .m file. The .m file is confirmed correct, but, we can not make our Simulink output match this.
>
> Is it the settings in the buffer? We have read the documentation for the buffer, but, are still left having difficulties.
>
> Thanks,
>
> Matt
>
> The link is http://rpi.edu/~brankd

It looks like you want all your (32761 vector) data into Simulink at the same time. It is easier to do this by loading the mat file into MATLAB workspace and then using the Constant block in Simulink. Enter the variable name in the "Constant Value" and make sure the value is a column vector. Also set the "Sampling Mode" in constant block to "Frame based" and set the Frame period based on your sample time. This will give you an output from the Constant block with a size 32768x1 which FFT block will be able to work on. To check the output of FFT use a "Vector Scope" from Signal Processing Blockset.

Hope this helps.
Navan