In my view, there are eight prime factors that determine what we hear from a loudspeaker, each governed by a wide – and interactive – range of electrical and mechanical parameters that come into play. These include the following:
• The materials used to construct cones and diaphragms (paper, aluminum, titanium, carbon fiber, composites, etc.).
• Low-frequency and mid-frequency cone geometry.
• The power, linearity, and type of magnetic circuit used in drivers (Alnico, ceramic, neodymium), as well as the voice coil wire (copper, aluminum, round, hexagonal, single layer, double layer, etc.), voice coil former material, acoustical reflections from the spider; the phase plug (in high-frequency drivers), and the concentricity of the voice coil in the magnetic gap.
• Horn material and flare rate.
• Enclosure material and construction quality (rigid is always better than poorly fitted joinery), as well as enclosure volume, bass port area, duct length, and port material (hint: wooden or PVC ports are better than cardboard and soft plastic).
• Composition of interior dampening material.
• Edge diffraction from the enclosure shape; the design of the crossover circuitry.
• Driver protection circuitry (if any).
• And much more!
This list makes it clear that a loudspeaker is the sum of its parts. Less apparent is the complex interrelationship of the many individual elements that must work together synergistically in order for a loudspeaker system to provide outstanding sound quality.
Factor 1: Frequency Response
There are two aspects to frequency response. First is the overall bandwidth of the response range. A wide bandwidth, say 40 Hz – 19 kHz, provides an immediate sensation of “high fidelity.” Conversely a narrow response range such as 200 Hz – 6 kHz will be perceived as “low fidelity,” though within that range the overall performance might be quite good indeed, such as a mid-range device in a 3-way system that was not designed to reproduce the full spectrum.
The second aspect is how even, or uneven, the response may be within the intended range. An even response equates to a flat loudspeaker…which usually is a good thing. When the response is uneven, the loudspeaker is not flat and can’t be relied upon for important judgments such as balancing input channels and setting EQ. While the overall bandwidth is a function of the loudspeaker’s design, an uneven response – if the magnitude is not excessive – can usually be corrected with precision parametric equalization, though this will require the use of a high-resolution spectrum analyzer.
Factor 2: Phase Response
Intimately linked with frequency response, phase response is quickly identifiable by using an FFT (Fast Fourier Transform) analyzer to characterize a loudspeaker. Every deviation in the frequency domain will yield a corresponding deviation in the phase domain.
Though we do not hear variations in phase response as readily as frequency response, such deviations are nonetheless present in virtually all real-world loudspeakers. When the other parameters have been optimized by careful design work, variations in phase response become quite audible.
What is the difference between phase, time smear, and group delay? They are three different terms that describe the same acoustical condition, that of time variations across a loudspeaker’s frequency range at a given point in space.
When drivers are not mechanically arranged so their acoustic centers are perfectly aligned throughout crossover, where they are both providing equal energy, one energy source will lag or lead the other in time. This can be partially corrected by incremental delay, but broadband delay may not solve the problem.
Each driver, as it approaches the extent of its response range, typically exhibits a deviation from a flat phase response all on its own – not just in relationship to the other driver. Fortunately, with modern DSP technology, phase filters and/or all-pass filters can be used to minimize phase versus frequency deviations.
What does “all-pass” mean? An all-pass filter alters time in relation to frequency, rather than altering frequency response like a parametric EQ. It provides delay, but as a function of frequency.
What’s the difference between time delay and phase delay? While the underlying mechanism is the same, when we speak of time delay, we’re usually referring to long periods of time, such as the differential between a main array and a delay tower.
When we speak of phase, we’re either talking about a 180-degree reversal (more accurately termed polarity), or we’re speaking of the relationship of arrival time in respect to frequency. There are almost no individual drivers that exhibit uniform phase versus frequency response when measured alone, and deviations are normally much greater when one driver interacts with another driver in a multi-driver system.
Where does group delay fit into all this? When phase response is linear (flat), group delay and phase delay are identical and are the same as time delay. In a non-linear system, group delay is the slope of the phase response at a given frequency. Variations in group delay cause signal distortion (not to be confused with harmonic distortion), just as deviations from linear phase also cause signal distortion.
Factor 3: Harmonic Distortion
This is extremely important because it determines much of what we perceive when we decide that we like one loudspeaker over another. All loudspeakers produce distortion, with most being three decimal points higher than any other device in the signal path – amplifiers included. The question is how much distortion, as well as how it varies as power levels vary, along with the nature of the distortion.
Let’s discuss how harmonic distortion is measured with an FFT. Typically, a low-distortion sine wave is applied to the loudspeaker. The acoustical response is then captured with a measurement grade microphone and viewed on an FFT. Ideally, the driver should produce only the fundamental frequency of the applied sine wave.
However, in the real world, the driver will inevitably produce second, third, fourth (and higher) harmonics that are easily seen on the FFT. The combined magnitude of all the harmonics is the THD, or Total Harmonic Distortion.
But there’s more. To fully understand a driver’s distortion characteristics, one has to change the frequency of the sine wave and look at the harmonics over a large range of frequencies and power levels, a time-consuming effort. What you’ll see is that the distortion products of most LF and HF drivers will increase as the frequency is lowered.
You’ll also see an increase as the power level is raised. In a high-grade driver this should be a linear function, i.e., 10 dB greater amplitude of the fundamental equals 10 dB greater amplitude of the harmonics.
At some point, however, as the driver is pushed hard enough, the harmonics will no longer maintain a linear relationship to the fundamental. It’s actually possible to measure a higher level of second or third harmonic distortion than that of the fundamental. In such case the driver is producing more than 100 percent distortion, and the sonic result is truly awful.
Knowing the frequency range and levels in which distortion starts to radically increase will greatly help when deriving low frequency port alignment, as well as determining optimal crossover points.
Factor 4: Non-Harmonic Distortion
This is even worse than the harmonic variety. When quality drivers are operated below their power limit, the distortion they produce is harmonically related to the fundamental. Apply a 100 Hz sine wave to a cone driver and the distortion “product” will consist of a 200 Hz component (second harmonic), a 300 Hz component (third harmonic) a 400 Hz component (fourth harmonic). and so on.
Though distortion is not desirable, harmonic distortion is, at least, related to music. The pure beauty of a fine piano might be compromised, but at least it will still sound like a piano. Not so with non-harmonic distortion.
When a loudspeaker’s distortion products are not related to the harmonic scale, the effect is a radical alteration in tonality. A piano might sound almost nothing like a piano, if the non-harmonic distortion products are high enough. Usually (but not always), non-harmonic distortion is the result of a mechanical problem, not a design issue, and can therefore be fixed.
Incidentally, when we say distortion “products,” we’re referring to the contribution of harmonic and non-harmonic energy that’s the product of a flawed transfer-function of electrical power being inaccurately converted to acoustical power.
We don’t want all of this extra energy coming out of our drivers, but it’s going to be there anyway. Driver designers minimize distortion by choosing optimal materials, while mix engineers can utilize the system well below its peak output power to keep distortion at very low levels, where it belongs. Distortion is not just related to output power, but is a function of output power, at least at this time of technological development.