A very useful routine that’s built into some analyzers for measuring distortion continually increments the drive level to the DUT (device under test) so that the operator can view the effect of level versus distortion versus frequency. This is important because measuring distortion at a single level or a single frequency won’t tell you much.
The distortion characteristics of an 800-watt woofer with 1 watt of power applied will not be the same as when 800 watts are applied. Nor will the harmonic distortion at 30 Hz be the same as at 100 Hz (hint: it will be significantly higher at 30 Hz and at 800 watts). The use of an analyzer will reveal these conditions without having to sacrifice your hearing by trying to determine them by ear.
Fighting Distortion
But what if you don’t happen to have an analyzer equipped with distortion measurement capabilities? Distortion can still be measured with an FFT (fast Fourier transform) or other narrow-band analyzer, but the process requires a bit more effort.
To make a harmonic distortion measurement, a low-distortion sine wave source is needed. If an analyzer doesn’t have a built-in adjustable-frequency sine source, an outboard generator can be used. Make sure that the sine wave generator is as pure and distortion-free as possible.
By applying a low distortion fixed sine wave to the DUT, and looking at the spectral content of the DUT, you can see the magnitude of the harmonics that are not present in the test signal. Since most loudspeakers produce one or two orders-of-magnitude greater distortion than most electronic equipment (amplifiers, pre-amps, etc), we’ll use a loudspeaker as an example. Figure 2 shows a typical screen commonly seen when viewing distortion on an FFT.
The fundamental tone from the generator is the tallest trace at the left of the screen. For this example, let’s say it’s set at 100 Hz. An ideal loudspeaker would reproduce only the 100 Hz tone and nothing more.
However, there are no perfect loudspeakers. So what is produced is a series of harmonics that are the loudspeaker’s distortion products. In other words, unwanted energy content that the loudspeaker is emitting but not being excited with.
As the drive level to the loudspeaker is increased, the fundamental will increase in amplitude. If the loudspeaker is linear, the harmonics will increase proportionately. As the level continues to increase, and the loudspeaker eventually becomes non-linear, the amplitude of the harmonics will increase at a rate that is greater than that of the fundamental.
Figure 3 depicts the distortion signature of a loudspeaker that’s become significantly non-linear due to the power applied. If even more power is applied, the second and third harmonics may even exceed the level of the fundamental, and/or the loudspeaker will be damaged.
In order to properly characterize a loudspeaker’s distortion across the full spectrum that it’s intended to reproduce, measurements must be taken at many different frequencies.
It’s better if they’re not octave-band related because the same problem (or lack thereof), usually stemming from some form of resonance, is likely to be repeated when the distortion products are harmonically related to a previous measurement.
In other words, once you’ve measured distortion at 100 Hz at various power levels, increase the frequency of the oscillator to 110 Hz, then 125 Hz, then 133 Hz, and so on, rather than jumping ahead to 200 Hz.
There is no hard and fast rule, but the greater the number of measurements taken, the better understanding you’ll have of the loudspeaker’s behavior. After spending a day or two doing this manually, it becomes quite evident that a distortion measurement routine, pre-programmed into a modern analyzer, can compress many hours of work into mere minutes.
