# Why is harmonic voltage flowing to the source

In non-linear loads, Current Harmonic Distortions are created, which in turn creates voltage harmonic distortion that are said to flow from the load to the source ( system ).

My understanding of Harmonic Distortions in AC circuits ( based on different papers as well as What exactly are harmonics and how do they "appear"? ) is that harmonics are not "created", but are merely a mathematical representation needed to do the Fourrier transform of the wave and explain it mathematically. If that is the case, how can these Harmonic Distortions flow back from the load ?

Any insights would be appreciated.

EDIT: Reading http://electrical-engineering-portal.com/res/Harmonic-detection-and-filtering.pdf, it seems I was mistaken and it is harmonic current that flows back to the source, causing a distorted line voltage in the load. My question then becomes, is this harmonic current flowing back through the neutral line ( 3-phase wye ) ? If so, how does this works in delta connected system ? Would this mean the current is flowing back to the source on the same line ?

• Your title says "harmonic voltage" and I have to say that I could guess what you mean but I think you should be totally clear here. – Andy aka Jan 12 '17 at 15:53
• Harmonic voltage might be a better way to say this. I am making the distinction between voltage and current harmonic, specifically relating to harmonic distortions. – m4mush Jan 12 '17 at 15:55
• I might argue that Fourier transforms are an invention to explain away the very real harmonic currents/voltages in a non-linear load. Who says that reality should have sinusoidal representation? – glen_geek Jan 12 '17 at 16:07
• @glen_geek, the advantage of sinusoids is that they are eigenfunctions of a linear system. If you input a sinusoid, you get a scaled version of that sinusoid out. If you input a square wave, for example, to a low-pass filter, you won't get a square wave out. – The Photon Jan 12 '17 at 17:28
• Given harmonics do not exist, you now have a freedom to engineer the system. With a radio at 313 MHz, will the 3rd harmonic of 104MHz MCU clock, at 312MHz, be a problem? – analogsystemsrf Mar 21 '17 at 2:56

Start with non-linear loads creating distortion. Very commonplace such as in a bridge rectifier. The capacitor gets charged only at the top of the rectified wave (see the black trace below): -

This results in an AC current that looks somewhat like this: -

Now this can be analysed and broken down into harmonics. Those harmonics are representations of the non sinusoidal shape of the waveform and as such, it would be daft to think of those harmonics individually being taken from the source but, it is reasonable to think of them harmonically when analyzing how these harmonics may get re-shaped as they pass down the cable.

how can these Harmonic Distortions flow back from the load ?

There is no difference physically between a signal that is created from summing harmonics AND the original signal that was harmonically analysed. So it doesn't matter which "model" you choose - both have the same influence back at the voltage source.

• Thank you @Andyaka, it does help me understand this a bit better. If I understand this, there is no such concept of harmonic voltage flowing back to the source. It is simply that sometime, the sum of the "peaks" in the harmonics can be much higher/lower than the "fundamental" voltage in the source -> load flow – m4mush Jan 12 '17 at 16:15
• I do have a follow-up question based on part of your anwser: – m4mush Jan 12 '17 at 16:32
• "both have the same influence back at the voltage source." : I do have a follow-up question based on that. How is the harmonic distortion affecting the voltage source ? Are you referring to the normal effects of voltage/current demand in AC circuit ( i.e. voltage drops on big motor starts without capacitors ) or to specific effects caused by the voltage distortion ? – m4mush Jan 12 '17 at 16:37
• @m4mush Keep in mind that Andy's scenario contains multiple sources. One supplies the input voltage/current sinusoid. The harmonic source is at the diode(s). It gets complicated fast. – glen_geek Jan 12 '17 at 17:25
• @m4mush I've probably generalized as much as I can on this so maybe if you come up with a focussed scenario it might help a bit more. You can develop this scenario here as an idea and I can give you a few pointers but it might be better to ask a new and more focussed question. – Andy aka Jan 12 '17 at 18:44