FFTs On Excel

messy

ny

It's not necessary to know everything about the Excel application to learn a lot about transforms.

Make up a function and take the transform

If the original time domain function was just sine curves with 0 phase angle, take the IMAGINARY() part of the transform to see what amplitude ends up in each frequency.

Make up something simple in the frequency domain, say just one frequency, and take the inverse.

To see the inverse transformed graph you may need to transfer the data to another column and add zero.

Bret Cahill

How about this:

Good luck

Brian Ancient and Modern Optics

this:

I've been trying to recover a signal, a polynomial, with match filtering on Excel but keep getting something that positively reeks of sinusoids.

OK, the "positively reeking of sinusoids" was just a really bad Fourier joke but I should get be able to recover the original polynomial made up of a lot of sine curves w/ phase angles or a lot of real and imaginary components..

Here's what do:

First make up a 2^n list of counting numbers in A0 - AN, the time domain. Then make some polynomial curve. Then add some not too very correlated noise. Then take the transform. The take the dot product with a reference that has been similarly transformed. Then take the inverse transform and the noise should be somewhat attenuated.

I don't want to see a sin or cosine curve if the original curve was a polynomial.

Bret Cahill

How do you expect to tell the difference?

What you ask for is similar to be given a number, say, 1, and determine in what context it originated: DSP? Physics? Economy? Schoolgrades? Counting socks in the drawer?

A number is just that, a number. A curve is a curve. And nothing more.

Rune

Normalized versions of the original and the final can appear together on the same graph.

Bret Cahill