[Phys-l] spectrum analyzer requires more than FFT

[John Denker <jsd@av8n.com>]

Josh Gates-fac wrote in part:

> - FFT frequency spectrum generation
...
> click-and-zoomable freq. spectrum analysis

Executive summary:  FFT is a low-level building block;  a spectrum
analyzer is a much higher-level thing.

That may sound like a mere wise-guy technicality.  A classmate tried
to explain it to me back in my larval stage;  I didn't understand or
believe what he was saying.  But he was right.  The distinction may
be quite helpful when shopping for software:  Assuming you want a
spectrum analyzer, shop for a spectrum analyzer!  If you just shop
for the "FFT" feature you are likely to get considerably less than
what you wanted.

Those who are not interested in details should skip the rest of this
note.

========================

I revised and extended my writeup on the practical details of
Fourier transforms and spectrum analysis
  http://www.av8n.com/physics/fourier-refined.htm
to explain in detail the distinctions I'm making.  There are some
example figures showing why a bog-standard FFT can be very misleading
as to the position, height, width, and symmetry of spectral features.

I also put up some  scilab  code to exhibit some of the lurid details,
showing some of what would be involved if you want to build a semi-decent
spectrum analyzer, using the FFT as a foundation-stone.
  http://www.av8n.com/physics/fourier-norm.sci

Obviously these practical details are not what 9th graders are expected
to worry about.  Building a spectrometer from scratch is more of a
grad-school exercise than a high-school exercise.  That's why the
teacher gets the big bucks:  he can acquire the appropriate thing for
the students to use.