4
\$\begingroup\$

I'm new to FFT and harmonic distortions. I'm dealing with a power system where the phase currents are distorted.

Now I don't understand the concept of FFT and why the harmonics seen in FFT relate to the current waveforms. FFT shows how the current waveform changes in gain by frequency instead of time. How is it possible for something to generate a waveform across all frequencies by just looking at a waveform from one frequency? Could someone explain this in a more intuitive way instead of using mathematical language? Also, if harmonics are the changes in gain for other frequencies, why do I care about them if I'm running at a single constant frequency? How can these changes from other frequencies affect something from another frequency?

Also, is an RLC band-stop filter a good way of cleaning up the waveform by simply blocking out the frequency range where the harmonics are? How would you block out multiple frequency ranges since there's more than one bad harmonic?

\$\endgroup\$
  • \$\begingroup\$ Got any links to what you've read so far? (oops sorry for the snarky comment.) \$\endgroup\$ – George Herold Jan 9 '15 at 1:32
  • \$\begingroup\$ en.wikipedia.org/wiki/Fast_Fourier_transform \$\endgroup\$ – rojas777 Jan 9 '15 at 2:32
  • \$\begingroup\$ It sounds like the fundamental thing you're missing is that you can represent most signals as the sum of multiple sinusoids at different frequencies. Even though your signal has a fundamental frequency, unless it's a perfect sine wave, it also contains a small amount of sine waves at other frequencies. \$\endgroup\$ – Greg d'Eon Jan 9 '15 at 11:14
11
\$\begingroup\$

They say a picture is worth a thousand words. And a handful of animated pictures...

enter image description here


enter image description here


enter image description here


enter image description here enter image description here


All of these from Wikipedia. You can see these in context in the articles for Fourier transform, square wave, and Fourier series.

As it turns out, any periodic waveform can be represented by a Fourier series, although to represent it exactly may require an infinitely long series. In engineering practice, however, above a certain point continuing the series adds no additional useful precision, and we can call a finite series "good enough".

Direct answers to your questions:

How is it possible for something to generate a waveform across all frequencies by just looking at a waveform from one frequency?

I think your mental hangup here is "one frequency". The point of Fourier analysis is that a square wave (say, a 60 Hz square wave) is not just one frequency. I mean it is, in the sense that it's just a 60 Hz square wave. But we can also decompose that square wave into a number of sinusoidal waves, each with their own amplitude and phase.

Also, if harmonics are the changes in gain for other frequencies, why do I care about them if I'm running at a single constant frequency?

Because, through the lens of Fourier analysis, you aren't actually at a single constant frequency unless your waveform is a completely undistorted sine wave.

How can these changes from other frequencies affect something from another frequency?

In a linear and time invariant system, they can't. This is formally described in LTI system theory. Each frequency component passes through the system completely independently. We can consider what happens for one frequency, then separately consider what happens for another frequency. Adding the two together, we get the same result as if the two frequencies passed through the system simultaneously. This makes Fourier analysis a very useful analytical tool.

Also, is an RLC band-stop filter a good way of cleaning up the waveform by simply blocking out the frequency range where the harmonics are? How would you block out multiple frequency ranges since there's more than one bad harmonic?

If the problematic harmonics are in just one band, yes. In a power system, it's likely that you will be dealing with 60 or 50 Hz with some degree of harmonic distortion, meaning it will also contain frequency components at integer multiples of the fundamental. If you can design a filter that allows that fundamental frequency to pass, but attenuates all higher harmonics, then no matter how distorted the input, the output will be a pure, undistorted sine wave. In a power system you might also care about the phase of voltage and current (power factor), and there are filters to fix that too.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.