THD vs Harmonics vs Total distortion

Firstly its late and im tinkering so this could be me being a complete nana

But I have noticed a strange anomoly

See the attached images..

This is the output of a DJ cd player playing a 24/48 1khz wav from a usb key

how can the thd graph show -87.7 at H5

BUT the THD is -94?

Its been a long challenging night but I though THD could not be lower than any single harmonic

I have noticed the main peaks in the main window graph do not align exactly with the harmonic frequency (i.e. the biggest peak other than the fundamental is at 4.899khz)

is this throwing it off? the distortion dbv (add measurement) seems to catch it

You can have a THD lower than a visible harmonic if your measurement bandwidth is insufficient to cover it.

What is your measurement bandwidth set at?

Hello Dean_OAV,

The signal at 4899 Hz is not a harmonic overtone of your fundamental frequency at 1000 Hz. Therefore it’s not considered, when calculating THD. Only integer multiples of the fundamental frequency (here 1000 Hz) are “harmonic distortion components”.

I do get all that.. but the THD Bar Graph Display IS picking up 4899hz as the 5th Harmonic..

yeah, also 2939 Hz, 6859, 10776 Hz, etc. are picked up as “harmonics”, very odd.

Did you miss to check “Round to eliminate leakage” in the signal generator settings?

Your FFT-window is quite wide. Try other FFT-windows, i.e. Hann.

Or an aliasing effect? Does it change when you increase your sample rate to 256k?

Hi @Dean_OAV

The bandwidths are different for the main THD computation and THD Bar Graph. I think this was a byproduct of changes to the THD computation at low frequencies a while back. But in short, the main THD computation uses a +/- 20 Hz bandwidth placed around the multiples of the computed peak frequency.

The bargraph is using a +/- 5% window, which what the main THD calc used to use. I’ll see about unifying these.

1 Like