Drivers and crossovers

You, dear reader, will probably immediately associate the term loudspeaker with those endearing final links in an audio chain. Correct, but in Spanish they are also known as acoustic screens -or simply screens-, by the name bafles (from the English baffle), a term accepted by the Royal Spanish Academy, simply as boxes -for short- and even as altavoces. But to avoid confusion, bear in mind that, strictly speaking, altavoz (driver) refers exclusively to those cone-shaped components installed in the loudspeaker cabinets. The job of a driver is to convert the electrical energy it receives into sound waves, which is why it is considered a transducer, since it transforms one type of energy into another.

We should point out that when we use the word driver in this article, we will be referring exclusively to one of those components, to distinguish it from the loudspeaker as a whole, a term we will also use. Our intention is to sketch out a few basic concepts about a complex subject full of technicalities and complicated formulas.

Broadly speaking, sound is the sensation produced in us by the vibrating motion of bodies transmitted through the air by waves of varying length. Frequency is the number of times these waves, or cycles, are repeated in one second, and its unit of measurement is the Hertz (after H. R. Hertz, the German physicist), whose symbol is Hz. In optimal physical condition, the human ear can perceive frequencies between 20 and 20,000 Hz, and there are musical instruments capable of generating notes (fundamental frequencies and their harmonics) that exceed the audible spectrum. A good loudspeaker must therefore be able to reproduce the widest possible frequency range, but the most important thing is that it should do so with maximum linearity (minimum deviation from a given reference axis) and the lowest possible distortion. The current state of technology does not allow the manufacture of a single driver capable of effectively combining these factors, although there are very good-quality drivers for particular uses. But they generally have severe limitations that make them unsuitable for strict high-fidelity sound.



Impossible with a single driver



To reproduce low frequencies, or bass, the driver has to displace a large amount of air, for which it needs a large-diameter cone (also called a diaphragm). At these frequencies, the driver's voice coil will easily move the cone back and forth, for example 80 times per second to generate an 80 Hz tone. But as the frequency, and therefore the speed of this back-and-forth movement, increases, the inertia of the mass of a large cone will prevent the voice coil from moving it with the speed and precision required for an 18 kHz tone, that is, 18,000 times per second.

Conversely, the small mass, and therefore low inertia, of a 2 cm diaphragm will allow its voice coil to move it quickly enough to reproduce frequencies even above 20,000 Hz. But this same driver will be completely unable to displace the amount of air needed to reproduce frequencies of 50 Hz. From this it follows that more than one driver is needed to build a loudspeaker with high-fidelity aspirations.

Given the great external similarity between a conventional driver and a coaxial one, we will briefly mention the latter. The coaxial driver has the particularity of consisting of two driver structures, each with its own voice coil and diaphragm, mounted concentrically, or coaxially, on the same chassis. To all intents and purposes they are two independent drivers, although they share a chassis and magnet. Each manufacturer uses its own technology and names it accordingly. Examples are KEF's Uni-Q and Tannoy's Dual Concentric coaxial drivers.

An important characteristic of every driver is its impedance, that is, the resistance its voice coil offers to the alternating current it receives from the amplifier. It is measured in Ohms, with standard values of between 4 and 8 ohms for hi-fi drivers. To group drivers, parallel, series and series-parallel connections can be used. In a parallel connection, the terminals of the same sign ("+" and

"-") are joined together separately, and the resulting terminals are taken to the amplifier. To connect them in series, the terminal of one sign on one driver is joined to the terminal of the opposite sign on the next, and so on, connecting the resulting terminals to the amplifier, as in the previous case. The series-parallel connection is a combination of the two.

With a parallel connection, the resulting impedance is divided as drivers are added, whereas with a series connection it adds up, so in some cases the result can be an impedance that is completely incompatible with any amplifier. Although this is a serious drawback, it is much more serious that all the drivers attempt to reproduce the full frequency range, thus generating interactions and distortion due to overlapping frequencies and to the inability of a driver to reproduce frequencies beyond its capabilities.



To each driver its own frequencies



In general terms, the typical frequency ranges of hi-fi drivers are usually 18 to 3,000 Hz for low-frequency drivers, 200 to 10,000 Hz for midrange drivers and 2,000 to 25,000 Hz for high-frequency or treble drivers. Even so, the usable ranges are much narrower. The response curve of any driver begins to roll off at both ends before reaching its limit, with the added problem that at the low end there is the undesirable phenomenon of the resonance frequency, a very pronounced peak in the response curve caused by a resonance of the driver's entire moving system. Below this peak, the response drops sharply. This phenomenon is particularly worrying in bass drivers, limiting the lowest frequency they can reproduce without distortion, so a good bass driver must have its resonance frequency as low as possible. However, it is unimportant in midrange and treble drivers, since with an appropriate filter the most suitable range can be selected, keeping well away from their resonance frequency.



The crossover



We are beginning to see the need to filter, at the electrical level, the range of frequencies each driver will receive. In loudspeaker construction, the range of frequencies to be reproduced is divided between two or more ways -or sections-, assigning to each the frequency range best suited to the drivers connected to it. Note that a way can have more than one driver associated with it; it is common to find three-way loudspeakers with five or more drivers, to give one example. This electrical division is carried out by crossovers, of which there are two types: active and passive. We will focus for now on the latter, which are always present in any loudspeaker with at least two ways. They are called passive because they are built almost exclusively with passive electronic components (active components are rarely used), such as capacitors, inductors and resistors, however complex the filter. They work on the basis of the capacitor's ability to let high frequencies pass through it while blocking low ones. Inductors have the opposite characteristic, letting low frequencies through and blocking high ones.

By combining capacitors with inductors and playing with their capacitance and inductance values respectively, different filters (ways) can be built which, together, form a crossover. Including one of these filters means that the amplifier will "see" as the resulting impedance of the loudspeaker that of each of its drivers separately.

Figure 1 shows the effect of a three-way crossover, in which the red curve corresponds to the low-frequency section, called a low-pass filter, which lets through all frequencies up to a given one, 500 Hz in the example. The green curve is the midrange section, or band-pass filter, which lets through all frequencies between two given ones, in this case 500 and 5,000 Hz.

The blue curve corresponds to the high-frequency section, called a high-pass filter, which lets through all frequencies above a given one, 5,000 Hz in the example. Note that the transition between the different ways is not abrupt; there is an attenuation slope determined by the number of capacitors and inductors in each section. This attenuation is measured in decibels (dB) per octave, an octave being a range of frequencies in which the highest is twice the lowest. The range between 50 and 100 Hz is one octave, as is that between 1,000 and 2,000 Hz. At the crossover point, the response is reduced by half (-3 dB), although in reality our ears will not detect it.



A complex crossover is not always better



With one inductor or capacitor per section, the attenuation slope will be 6 dB per octave, a slope that increases by another 6 dB for each capacitor or inductor added. Filters with a 6 dB slope (first-order) are the simplest and those that produce the fewest phase errors. They normally consist of one component, an inductor or capacitor per section, connected in series with the respective driver. In second-order filters, the attenuation slope is 12 dB per octave, and they consist of two components per section, one connected in series with the driver and the second in parallel. They are often used for the treble way, despite the phase shifts they generate, which can sometimes be corrected by reversing the polarity of the driver's connection. Third- and fourth-order filters, with attenuation slopes of 18 and 24 dB, use three and four components respectively. When it is necessary to reduce the energy a driver receives or to fine-tune the cutoff frequency, resistor networks are used.

Figure 2 shows the theoretical schematic of a three-way crossover with 6 dB attenuation slopes for the low and mid frequencies and 12 dB for the high frequencies. Note in the latter section a capacitor in series and an inductor in parallel with the driver.

One might think that the steeper the attenuation slope, the more perfect the separation between frequency bands and therefore the closer we will be to the theoretical ideal. On the contrary, piling up components can compromise signal quality, producing serious phase shifts between the ways that are far from easy to control, especially in even-order filters. In addition, a loudspeaker with a complex crossover is a strong candidate to become a serious challenge for many amplifiers on the market. The current trend is to build crossover networks with few components but of very high quality, such as air-core inductors, which require a greater number of turns, resulting in large, expensive inductors for high inductance values, non-polarized capacitors made of polyester or rare resins, and metal film resistors. Filter design uses mathematical functions defined by scientists, hence the names Butterworth, Bessel, Linkwitz filters, etc.