FFt phases: software for a neophyte?

I'm a biologist and I'm trying to understand why a particular species of
insect responds to some temporal patterns in pulse trains but not to
others.  I've identified 2 peaks in the FFt that must be present and of
similar magnitude to generate a strong response from the insects; if a
stimulus is missing one peak, or if the second peak is lower than the
first, the response of the animal drops.  My problem is that I have a few
outliers in which signals were unattractive despite having the correct
frequency attributes described above.  I'm wondering whether the phases of
the two critical frequency components might be important.  The bioacoustic
software I'm using (Bat Sound) provides neither a phase plot nor the real
and imaginary numbers I'd need to calculate the phases myself.  Can anyone
suggest some simple-to-use software that will either provide the phase
information, or provide the numbers needed to calculate the phases? I have
no background in engineering/computer science and don't know how to use
Matlab. Thanks for your help.

On 20 Feb, 21:57, "SarahB" <bus...@missouri.edu> wrote:
> I'm a biologist and I'm trying to understand why a particular species of
> insect responds to some temporal patterns in pulse trains but not to
> others. &#4294967295;I've identified 2 peaks in the FFt that must be present and of
> similar magnitude to generate a strong response from the insects; if a
> stimulus is missing one peak, or if the second peak is lower than the
> first, the response of the animal drops. &#4294967295;My problem is that I have a few
> outliers in which signals were unattractive despite having the correct
> frequency attributes described above. &#4294967295;I'm wondering whether the phases of
> the two critical frequency components might be important. &#4294967295;The bioacoustic
> software I'm using (Bat Sound) provides neither a phase plot nor the real
> and imaginary numbers I'd need to calculate the phases myself. &#4294967295;Can anyone
> suggest some simple-to-use software that will either provide the phase
> information, or provide the numbers needed to calculate the phases? I have
> no background in engineering/computer science and don't know how to use
> Matlab. Thanks for your help.

Given the nature of your problem and your stated background
and knowledge of computers, I would suggest that USENET might
not be the best place to search for help. You might find it
more productive to find somebody who has the necessary skills
with data analysis and computers and who can work along side
with you on a daily basis. Your email address indicates you
are with some university, so I would suggest that you try to
find somebody in the maths statistics department who works
with time series analysis, or somebody in the electrical
engineering department who works with DSP.

Rune

On Feb 20, 12:57 pm, "SarahB" <bus...@missouri.edu> wrote:
> I'm a biologist and I'm trying to understand why a particular species of
> insect responds to some temporal patterns in pulse trains but not to
> others.  I've identified 2 peaks in the FFt that must be present and of
> similar magnitude to generate a strong response from the insects; if a
> stimulus is missing one peak, or if the second peak is lower than the
> first, the response of the animal drops.  My problem is that I have a few
> outliers in which signals were unattractive despite having the correct
> frequency attributes described above.  I'm wondering whether the phases of
> the two critical frequency components might be important.  The bioacoustic
> software I'm using (Bat Sound) provides neither a phase plot nor the real
> and imaginary numbers I'd need to calculate the phases myself.  Can anyone
> suggest some simple-to-use software that will either provide the phase
> information, or provide the numbers needed to calculate the phases? I have
> no background in engineering/computer science and don't know how to use
> Matlab. Thanks for your help.

First some questions: Are the two frequency components
of interest related by an integer ratio?  Or are the
initiations of the two pulse trains synchronized to each
other in some manner and of finite duration?

SarahB wrote:
> I'm a biologist and I'm trying to understand why a particular species of
> insect responds to some temporal patterns in pulse trains but not to
> others.  I've identified 2 peaks in the FFt that must be present and of
> similar magnitude to generate a strong response from the insects; if a
> stimulus is missing one peak, or if the second peak is lower than the
> first, the response of the animal drops.  My problem is that I have a few
> outliers in which signals were unattractive despite having the correct
> frequency attributes described above.  I'm wondering whether the phases of
> the two critical frequency components might be important.  The bioacoustic
> software I'm using (Bat Sound) provides neither a phase plot nor the real
> and imaginary numbers I'd need to calculate the phases myself.  Can anyone
> suggest some simple-to-use software that will either provide the phase
> information, or provide the numbers needed to calculate the phases? I have
> no background in engineering/computer science and don't know how to use
> Matlab. Thanks for your help.
> 
> 

I note that Mizzou has both "Computer Science" and "Electrical & 
Computer Engineering" degree programs so Rune suggestion is good.

I would also suggest looking at _The Scientist and Engineer's Guide to 
Digital Signal Processing_ ( www.dspguide.com ) for some general 
background. One advantage you'll find is if you jump into it in the 
middle and your missing particular background he frequently tells you 
where in the book to go.

For your problem, you will want to look at "Chapter 12: The Fast Fourier 
Transform" ( www.dspguide.com/ch12.htm ), specifically "FFT Programs"
( www.dspguide.com/ch12/3.htm ). It gives stand alone BASIC code with 
the real and imaginary components in separate arrays to ease presenting 
phase data as you wish.

I'm not familiar with Matlab, but use Scilab ( www.scilab.org ), a free 
alternative. Feel free to ask questions if you get stuck with 
cross-discipline communication problems. My email is valid.

Hi Sarah,
Also, there is prewritten FFT program on my website that reads a text
file, calculates the FFT, and writes the results to a text file.  You
could
view the resulting input and output in Excel.  Run the program for
instructions.  Feel free to e-mail me if you have problems getting it
running.
Regards,
Steve

http://www.dspguide.com/download.htm

Yes, the higher frequency peak is the first harmonic of the lower frequency
peak .  It's actually a single train of pulses, but each pulse contains two
rise-times: the first rise-time takes the pulse to 75% amplitude, then
there's a plateau before the amplitude rises to 100% (followed by a second
plateau and a single fall-time).  I'm going to look into the suggestions
made by the others.  Thanks for the help.

SteveSmith wrote:
> Hi Sarah,
> Also, there is prewritten FFT program on my website that reads a text
> file, calculates the FFT, and writes the results to a text file.  You
> could view the resulting input and output in Excel.

Now that would be a bit roundabout. If you already use Excel, you
might as well calculate the FFT in Excel.

Regards,
Andor

Hi Andor,
You can calculate the FFT in Excel?  Cool-- I didn't know that!
Regards,
Steve

SteveSmith wrote:
> Hi Andor,
> You can calculate the FFT in Excel? �Cool-- I didn't know that!

I have a German version of Excel 2000 on this computer. I found the
FFT function under "Extras -> Analyse Funktionen -> Fourieranalyse".

Excel can even handle complex numbers, using a bunch of functions with
bizarre names ... For example "imabs()" computes the absolute value of
a complex number, whereas "abs()" computes the absolute value of a
real number.

Regards,
Andor

Thanks a lot-- very good to know.
Steve