Zaph|Audio

There are two general types of harmonic distortion tests - Single frequency spectrum and fixed harmonic sweeps. I do sweeps, while many others do single frequency.

Single frequency has two primary benefits. First, the level at the fundamental frequency can be adjusted to the same level as other drivers tested, making easier a direct comparison at a single frequency. Second, very tall order harmonics can be seen. For example all the harmonics above F6 can be seen, along with other spectrum noise.

Fixed harmonic sweeps have several benefits. The first and foremost is that distortion is shown for all frequencies at once, rather a single selected frequency. Single frequency testing requires *many* tests to get a complete picture of distortion across the spectrum, and even then, sharp peaks in distortion could hide between 2 tested frequencies. With a sweep however, the primary harmonics F2 through F5 will be clearly shown at every frequency.

For example, if speaker with a large and very audible peak in the 3rd harmonic at 1.5kHz had single frequency tests done at 1kHz and 2kHz, the harmonic distortion would be completely missed. This is particularly important when testing stiff drivers with breakup nodes within the audible band. This is the primary reason I choose sweeps for harmonic distortion testing. There is a lot more information in one image.

We can't have everything with sweeps, and we loose the ability to see harmonics above F5. In my opinion, this is a fair trade off. Taller order harmonics far above the fundamental frequency are more audible than a low order harmonic of the same level. But tall order harmonics are typically at a much lower level. Looking at the data with trending in mind will help see around these limitations.

Reading harmonic distortion sweeps is fairly easy. This section will deal with interpeting the data and "reading between the lines" to help see around the limitations of this type of testing.

These sweeps are not EQ'ed to flat. Nor are they filtered or crossed over in any way. This is both an advantage and a disadvantage depending on how the driver is used. It is an advantage if the frequency in question falls within a range that is intended to be run without any sort of response shaping or crossover. For example, the top two octaves of a tweeter or the midbass of a woofer. In that case, what you see is what you get. It is a disadvantage when the response curve will be modified to reach a target response shape. For example, an extended low end of a tweeter will have it's harmonic distortion reduced in the area where the crossover begins to take effect. A woofer crossed over to a sub will have it's audible distortion reduced as the woofer rolls off.

Related to the above, the enclosure tuning for woofers affects the low end distortion. The baffle dimensions and driver location for all drivers will also affect the distortion level due to baffle diffraction ripple response and baffle step response due to 2pi to 4 pi conversion. The most important thing to remember when looking at these harmonic distortion sweeps is how the drivers will be used.

There's a lot more information in these F2-F5 sweeps than may first meet the eye. You can get a pretty good idea of what's going on further up in the spectrum by looking at the level of the F4 and F5 in relation to F2 and F3 respectively. This is a form of trending. For example, to take a guess at tall odd order harmonics, draw an imaginary curve from the fundamental to the F3 and F5, and it's almost guarenteed that the F7 and F9 are going to fall close to where that curve extends.

For intermodulation distortion, there is nothing more definitive than looking at a 2 or 3-tone spectrum, but in my experience an F4 or F5 of a relatively high level almost certainly points to high IMD as much as it points to high spectrum noise. While I do not provide IMD plots, be aware that drivers with high HD are probably going to have high IMD and high spectrum noise also. All these forms of distortion are closely related.