# Current harmonics via FFT — do peaks matter?

Typically, one would perform an FFT on a current waveform and measure it's magnitude at multiples of the line frequency (e.g. 50Hz, 60Hz) to get the harmonics, and subsequently the Total Harmonic Distortion.

• If I know that my line frequency should be 50Hz but the FFT peak is not exactly at 50Hz (say, it is 55Hz), then what is my fundamental? 50Hz or 55Hz?

• How about the harmonics? Should I then measure at multiples of 50Hz or 55Hz?

• If say my fundamental peak is exactly 50Hz, but my second FFT peak is at 110Hz, where should I measure the 2nd harmonic? 100Hz or 110Hz? In other words, should I seek out the peaks?

• Lastly, THD is usually defined to take the sum at multiples of the fundamental. What if my FFT peaks are not exactly at multiples of the fundamental?

Update: - I was trying to measure THD for my power supply with line frequency of 50Hz. If measure the peak of the FFT, it's 50.35Hz. If I record that as the fundamental, the fractional points add up in the subsequent harmonics (e.g. for the 29th harmonic, the value becomes 1460.26 vs 1450 if 50Hz was used as the fundamental). From the FFT plot, clearly 1450 is the peak, while 1460 is not. - Should I look for peaks at each harmonic?

• Please tell us about your 1-Sampling Rate $Fs$ and 2-Size $N$ of FFT – Fat32 May 24 '16 at 11:33
• "What if my FFT peaks are not exactly at multiples of the fundamental?" then you are most likely doing something wrong, or if these are measurements, your equipment is imprecise (non-constant group velocity?). – oystein May 24 '16 at 14:24
• @Fat32 Fs=1.17Hz, N=64k. Actually, this is the oscilloscope's FFT waveform. – Ryuu May 25 '16 at 5:06