Prev: imrect
Next: Ant colony algorithm (ACO)
From: Skip Williams on 19 May 2010 20:50
Looking for algorithm to convert an fft magnitude vs frequency to compare with another fft of different fmax and different number of lines.
From: EE Student on 19 May 2010 21:05
"Skip Williams" <swillcac(a)aol.com> wrote in message <ht20vv$jpn$1(a)fred.mathworks.com>...
> Looking for algorithm to convert an fft magnitude vs frequency to compare with another fft of different fmax and different number of lines.


Can you please expand on what you are trying to do, this is difficult to understand, possibly post some code that you have tried and point out where you are stuck
From: TideMan on 19 May 2010 21:12
On May 20, 12:50 pm, "Skip Williams" <swill...(a)aol.com> wrote:
> Looking for algorithm to convert an fft magnitude vs frequency to compare with another fft of different fmax and different number of lines.

You can't.
FFT produces discreet data at particular frequencies. The data are
not continuous.
Interpolation between those is wrong, unless you smooth in the
frequency domain to approximate a continuous function.
You need to recalculate the 2nd set of data using the same FFT
parameters that were used for the first.
From: Skip Williams on 19 May 2010 22:20
"EE Student " <np7(a)cec.wustl.edu> wrote in message <ht21se$ia8$1(a)fred.mathworks.com>...
> "Skip Williams" <swillcac(a)aol.com> wrote in message <ht20vv$jpn$1(a)fred.mathworks.com>...
> > Looking for algorithm to convert an fft magnitude vs frequency to compare with another fft of different fmax and different number of lines.
>
>
> Can you please expand on what you are trying to do, this is difficult to understand, possibly post some code that you have tried and point out where you are stuck

Sample rate conversion should be the topic. Take a fft magnitude spectrum sampled at 2400 hz and convert it (approximately) to a magnitude spectrum sampled at 2000 hz.
From: TideMan on 19 May 2010 22:36
On May 20, 2:20 pm, "Skip Williams" <swill...(a)aol.com> wrote:
> "EE Student " <n...(a)cec.wustl.edu> wrote in message <ht21se$ia...(a)fred.mathworks.com>...
>
> > "Skip Williams" <swill...(a)aol.com> wrote in message <ht20vv$jp...(a)fred.mathworks.com>...
> > > Looking for algorithm to convert an fft magnitude vs frequency to compare with another fft of different fmax and different number of lines.
>
> > Can you please expand on what you are trying to do, this is difficult to understand, possibly post some code that you have tried and point out where you are stuck
>
> Sample rate conversion should be the topic.  Take a fft magnitude spectrum sampled at 2400 hz and convert it (approximately) to a magnitude spectrum sampled at 2000 hz.

If the two time series are EXACTLY the same duration, the frequencies
will be the same up to 1000 HZ (Nyquist), so you can compare directly.
This is because the interval in frequency is simply the inverse of the
duration:
df=1/T;
 | 
Pages: 1
Prev: imrect
Next: Ant colony algorithm (ACO)