translation: Raul Sanchez

History and development of acoustic systems.

What's in the past?

Modern speakers, as disgusting as they are, are still significantly BETTER than MOST speakers from the 50's. Very few of the hi-fi luminaries owned huge speakers like the Altec “Voice Of The Theater” A-7,

or Bozak B-305

15" Tannoy or Klipsch horns.

The average audio enthusiast had to make do with something like 12-inch University, Jensen or Electro-Voice coaxial heads built into huge resonating boxes made of glued plywood, with a single layer of glass wool on the back wall. The FI port was a large hole in the housing, which led to hum being mixed into the sound (since the ports were tuned to too high a frequency) with an increase in the frequency response by 6-12 dB in the region of 80-150 Hz. Have you ever heard of a refurbished jukebox?

Coaxial or, even worse, triaxial heads were characterized by paper resonant kinks at frequencies of the order of 300 Hz and above, cavity resonances (thank you to the horn elements glued into the middle of the speaker) starting at frequencies of 800 Hz and above, horn resonances throughout the entire operating range of the short horn, as well as resonant breaks of phenolic diaphragms, starting from 8 kHz and above. A “decent” head of this type, as a rule, had an uneven frequency response from +/-4 to +/-8 dB, and even in order to achieve such indicators, it was necessary to dampen it to the “I can’t” level.

So, such speakers of the early hi-fi period acquired the reputation of “boom-boom-chirIk-chirIk” quite deservedly. The sound quality was more reminiscent of an old neighborhood theater or the Central Park of Culture and Culture named after. Gorky than the sound of modern acoustics. Tube electronics did their best to help SWEETEN the roughness of the sound, but this still did not save the terrible quality of the speakers of that time. Without a doubt, the first generation speakers from Quad, RCA LC-1A, Tannoy and Lowther can compare with modern speakers, BUT... at that time there were so FEW of them, they were so RARE, and so EXPENSIVE that it is better not to remember. Well, how expensive? Classic speakers cost about the same as a NEW Volkswagen or a down payment for a new apartment!

It is quite obvious that the developers of that time did not have solid knowledge of how to model or predict the behavior of speakers in the low-frequency region, and the materials used to make tweeters do not stand up to any criticism compared to modern ones. Today, accurate, precise bass is taken for granted, and modern tweeters are simply excellent.

This is where the problems with modern acoustics begin, in the mid-frequency region, which is not “given” to computer development tools as well as the bass or high-frequency region. The “zest” and dynamism of the best CLASSIC SPEAKERS is in the midrange – the most important and most complex part of the entire spectrum! Progress in the midrange sector is very slow for various reasons. Where hearing reaches maximum sensitivity, the heads begin to operate at the border of their design frequency range, and the designer has to SIMULTANEOUSLY fight for spectrum flatness, polar response (dispersion), intermodulation distortion, impulse response, suppression of delayed resonances, diffraction at acoustic design angles, and for “unifying” the opposite characteristics of transition filters.

In the late 60s, large 12- and 15-inch FI systems were replaced by AR, KLH, Advent speakers and other small bookshelf speakers of the 60s and 70s. The new acoustics had 8-inch bass drivers, heavily “draped” with felt, small closed cabinets, phenolic dome tweeters, minimalist crossovers and extremely low sensitivity. By modern standards, they were dull, very dull, with a mediocre stereo panorama and rough sound, characterized by low resolution. All this was due to the use of minimalist crossovers, undamped standing waves in the acoustic design, NOT using an axisymmetric arrangement of heads, as well as problems with diffraction on decorative corners of drawers, grill subframes and the “heavy” matter of the permanent fabric meshes themselves.

And although the new bookshelf speakers were characterized by smoother frequency responses, measured using simple measurement methods of that time, the wonderful “zest” and “liveness” of the best examples of speakers of the 50s was lost. Only in the late 1980s did speakers with high sensitivity and new diffusers reappear, as well as new powerful measurement systems. During the period from the late 60s to the late 80s, ACCURACY of reproduction and NEUTRALITY of sound were put at the forefront.

It is quite interesting that when comparing transistor amplifiers the first generation like the Dyna 120, Crown DC-300 or Phase Linear 400 with the classic tube Dyna Stereo 70 or Marantz 9 were preferred first (even by J. Gordon Holt himself of Stereofile!). This says a lot about the resolution of, mind you, the BEST speakers of that time. Progressive improvements in speaker design over the past decades now reveal the REAL sound quality of those first generation transistor amplifiers - which is, in fact, terrible. But the sound of “refreshed” classic tube amplifiers, on the contrary, is the same or even better than the sound of the most expensive transistor amplifiers of our time.

Here is an incomplete list of problems that modern speaker developers have to face:

1.
It is fundamentally impossible for two-speaker STEREO to recreate the original acoustic wavefront - instead, it creates a phased, weakly realistic image of a small size, which, after prolonged listening, causes fatigue for many listeners (especially for NON-audiophiles). This virtual image is highly unstable and highly dependent on the location of the listener, the spectral distribution of energy and the properties of the listening room.

Even a simple center mono image has been shown to suffer from terrible comb filtering in the 1 to 4 kHz range, which is why lead vocals sound completely different when played through a single speaker or stereo pair. Psychoacoustic studies show that with a TWO-channel playback system, at least 3 speakers are required to more or less accurately recreate the tonal quality of centrally positioned sound images, such as vocalists.

2.
Large amounts of harmonic, intermodulation and cross-modulation distortion are superimposed on the mechanical resonances of the heads, resulting in spectral energy being concentrated at individual frequencies. Various head damping technologies, as a rule, improve the spectral characteristics (frequency response in any case), but they do not provide much improvement in terms of eliminating inflection modes, so the distortion spreads over a much wider frequency range.

The narrow-band nature of resonant distortion in speakers is the reason why measurements of harmonic or intermodulation distortion at any particular frequency are completely useless. To obtain a useful frequency dependence of distortion, an expensive measurement system with a tracking generator is required. It is then possible to obtain very different spectra for, for example, the second and third harmonics, as well as curves resembling a topographic map of the seabed. Any questions about “average” distortions, in their absurdity, resemble the question “What is the average depth of the Atlantic Ocean?”

To avoid resonant distortion, the head diaphragm must have a density close to that of air, and be characterized by absolutely uniform acceleration of the ENTIRE surface at ALL frequencies. Until now, man has NOT been able to create anything like this. As a consequence, ALL speakers exhibit tonal coloration ranging from slight to severe. Moreover, some types of coloration are constantly present, while others appear only at increased or decreased volume levels. The listener's musical PREFERENCES can easily mask the presence of these problems, especially if the listener listens to music with a relatively SIMPLE spectral composition (for example, Jazz).

3.
The energy of standing resonant waves accumulates in the heads (with the exception of plasma emitters with “zero” mass), acoustic design and, in fact, in the KdP itself. Unwanted mechanical energy must be quickly “merged” in two ways: by rigid, low-loss mechanical connections to the Earth (hard path from magnet to stand, from stand to floor, from floor to Earth) and dissipated as heat in high-loss amorphous materials such as such as lead, sand, sorbothane, etc. Energy that is not dissipated is re-radiated as background noise by EVERY mechanical part of the head or body, each of which in turn has its own resonant signature.

In any REAL speaker, regardless of the principle of its construction, at any moment in time there are hundreds of resonances in the form of standing waves, which are realized in a time ranging from milliseconds to several seconds. These resonances constantly superimpose on the structure of the music, change the tonal coloring, distort and mask the reverberant qualities of the original recording, and also “flatten” and blur the stereo panorama.

In speakers with a perfectly flat frequency response, this type of “hidden” resonance is the MAIN source of coloration. This is also the reason why one-third octave pink noise measurement methods have been replaced by the much more revealing technologies of TDS, FFT, MLS.

4.
The radiation pattern changes dramatically with frequency, and especially sharply in transition zones. In addition, the radiation pattern is further distorted by diffraction re-emission on EVERY sharp edge of the housing (regardless of size or type - this also applies to tiny satellites and planar speakers).

Diffraction that occurs on ANY sharp edge of the housing is a source of delayed phantom antiphase radiation that interferes with direct sound from the heads. These secondary phantom “emitters” are the cause of noticeable ripples in the frequency response in the midrange region (up to 6 dB) and sources of “delayed” sound that violate the timing relationships so necessary for the correct creation of a stereo image. These dispersion problems are audibly perceived as CD-dependent coloration, rough midrange, blurry stereo, and sound sticking to the speakers.

This list covers only part of the AS problems. There are other problems - though not so serious - which, nevertheless, you can learn to notice with some training. These problems arise in ABSOLUTELY ALL speakers - be it based on dynamic emitters, horns, ribbon emitters, electromagnetic planars, electrostatic planars and all the others. All these speakers are characterized by a large amount of harmonic and intermodulation distortion, concentrated at certain frequencies. All of these speakers store and emit significant amounts of resonant energy, and they all have a strong frequency dependence of dispersion, which is further aggravated by diffraction re-emission.

This is why ANY tales of “perfection” or “colossal breakthrough” should be taken with GREAT SKEPTICISM. At least some product of miracle technologies is free from the above-mentioned disadvantages? NO. The only thing that has really improved recently is the MATERIALS used, as well as the technology of MEASUREMENTS and computer SIMULATION. That's all.

Since ALL acoustics, WITHOUT exception, have SERIOUS shortcomings, albeit in the ABSOLUTE sense of the word, it depends on the consumer which acoustics he will choose and what he will demand from it. “Perfect sound forever” is just a stupid marketing slogan, the essence of which is absolutely NOT achievable for either an artist or an engineer. First of all, because the materials that would make it possible to create something like this simply do NOT exist.

Main “schools” of acoustics development

Since ALL developers are forced to choose based on SUBJECTIVE (or marketing) considerations, there is simply NO ONE Right or Wrong way to develop speakers. If someone tells you that this is not so, then we can only advise you to examine in more detail the “personal biases” of the declarant in order to make sure that if he is not an adherent of religious fundamentalism, then he is certainly inclined towards “rational-scientific” fundamentalism.

What is my personal position? I am definitely NOT an audio fundamentalist, or any kind of fundamentalist at all.

I really respect the flatness of the frequency response, minimal intermodulation distortion, the smallest possible number of delayed resonances (waterfall purity) and low diffraction. Moreover, I am looking for that elusive quality of sound that can be called “the bloom of life” or “reach out and touch”, i.e. a feeling of physical involvement in the performance. For those of you who have never experienced such sensations, all I can say is that it occurs in nature, but is about as RARE as a perfect double rainbow.

In the next chapter, I'll tell you about the different paths that designers are forced to follow to achieve sonic perfection.
Smooth frequency response – “school” of OBJECTIVE development

Most British and Canadian speakers fall into this category. They are characterized by smooth frequency response. At the same time, the British “school” of the BBC attaches the greatest importance to the AXIAL frequency response, taken from a distance of 2 meters, in combination with minimizing delayed resonances, and the Canadian “school” of the National Research Committee pays the greatest attention to the frequency response averaged over the hemisphere “looking” at the listener. These development priorities were set respectively by BBC broadcasting professionals and National Research Canada hearing experts.

This “school” of development is most consistent with the philosophy of OBJECTIVE ENGINEERING. It is no coincidence that engineers with master's degrees or doctorates in acoustics design speakers following THIS philosophy. These engineers cannot take seriously the "special" wiring cables, audiophile resistors, capacitors, and other MYSTICS of the direct heated triode series, nor anything else that is IMPOSSIBLE to be RELIABLY and REPEATEDLY heard in double blind testing.

D.I.L. Shorter from the BBC was the first who, in the late 50s, ACCURATELY measured and IDENTIFIED the sources of resonances of heads and cabinets, and many English-made speakers, as if in confirmation of this philosophy, are to this day characterized by excellent characteristics in this area. Since resonances can be heard even if their level is 20dB below the level of the desired signal, the BBC became the FIRST organization to identify and measure the COLOR of the sound, which is not “visible” at all on traditional frequency responses.

It took American developers as many as 20 years to RECOGNIZE the importance of this “hidden” color, and the real breakthrough in this area occurred when in the early 70s Richard Geyser invented time delay spectrometry (the so-called TDS technique), which found its first embodiment in measuring TEF system manufactured by Techron. 10 years later, Lipschitz and Vanderkoy invented the Maximum Length Sequence Analysis System, which was commercialized as a computer ISA card manufactured by DRA Laboratories. Over the course of 30 years, measurements of delayed resonances have evolved from specialized measurement instruments used ONLY within the BBC, i.e. from the $150,000 Hewlett-Packard custom FFT minicomputer used in the KEF labs, through the $12,000 TEF system manufactured by Crown, to the $3,500 generic MLSSA board that can be plugged into ANY PC.

MLSSA is currently the INDUSTRY STANDARD for all major speaker manufacturers. If you are primarily interested in the frequency response and are not very concerned about the mystery of interpreting transient characteristics and graphs of the ECG (waterfalls), then LMS products costing only 1000 bucks will be a good choice. LMS is widely used for product quality control because it can be easily configured for pass/fail mode in terms of frequency response. Another interesting product is CLIO, which can be compared to MLSSA in terms of functionality and has the bonus of 16-bit precision and low cost ($1600 including microphone).

In recent years, a lot of software, which is designed to use high-end sound cards, which further reduces costs to about $600. The only problem with most audio cards is the maximum sampling frequency of 44.1 kHz. In order to ACCURATELY measure the impulse response of a tweeter, the anti-aliasing filter located in front of the DAC must have a relatively flat slope (Bessel or Butterworth), and this, in turn, requires a sampling frequency of 96 kHz or higher. If the manufacturer sound card answers the question about the maximum sampling frequency or the slope of the anti-aliasing filter EVALUATIVELY, it is better to REFRAIN from purchasing such a card.
Let's take a break from the “schools” of speaker construction for a while and learn how to choose speakers.

First of all, DO NOT fall for MARKETING gimmicks! Remember: EVERYTHING, i.e. ABSOLUTELY ALL heads have their own sound signature, which, at best, can be kept in check by equalizing in the crossover, but it is IMPOSSIBLE to completely get rid of it. And although equalization in the crossover may well “straighten” the frequency and time characteristics of the head, there is no escape from intermodulation distortions that arise from resonances in the diaphragm or suspension. ALL physical materials have resonant modes, so if the head is made of physical materials, then it will DEFINITELY have resonant modes, no matter how cracked.

Since we all have to work with imperfect materials, here are a few notes that should help us come down from the “sky” of abstract ideas and concepts to Earth and begin to enjoy the sound.

1.
Aim for speakers with a SMOOTH and even frequency response and the most correct impulse response possible (Tecnics SB-8000 remember?). Take steps to eliminate any REFLECTIONS from the FACE of the speakers (remember Dunlavy?), as well as from possible interior panels (remember the terrible resonance in the Oscar Hale Kitara acoustics?). Reflections are audible much stronger and much more unpleasant than one might assume by looking at the bumps they create on the frequency response. The inside of the case should be carefully covered with FELT (85% wool) - this will ensure good suppression of reflections. As for the exterior, heads should NEVER be placed in any kind of cavity or recess, as even the best felt absorbers have very modest absorption capacity. The best thing is if the head is mounted strictly flush with the front panel, and the EDGES of the case are ROUNDED (remember the Technics SB-RX50?). Resonances from the drivers and acoustic design must be reduced to levels where they are virtually unnoticeable. Speaker enclosures must, first of all, be very rigid (remember the Yamaha NS-1000 Monitor?). Damping is a SECOND level task. To strengthen the body, glued plywood 2 centimeters thick is quite suitable.

2.
Avoid crossovers with transition frequencies falling in the most CRITICAL range for hearing, from 300Hz to 3kHz (Yamaha NS-1000 Monitor has a midrange head that operates in the band 500Hz-6kHz!!!). The phone companies were not wrong when they chose this band as the most significant part of the spectrum! This frequency region must be absolutely flawless, characterized by a FLAT frequency response and very low distortion. Even if a crossover is PERFECTLY tuned and executed, its presence is still SLIGHTLY NOTICEABLE. That is why it would be good to “take him out” of the critical zone.

3.
A well-controlled, peak-free crossover region is extremely IMPORTANT. If you can achieve this, in the transition zones you automatically get a smooth phase response and frequency response, and in general, a very high-quality “middle” and a serious increase in the depth of the stage.

4.
Well, and finally, it is necessary to reduce INTERMODULATION distortion as much as possible throughout the entire audible range, and this is ESPECIALLY true in the region from 500Hz to 5kHz. And this means selecting heads with well-designed magnetic systems, as well as using midrange drivers and tweeters that are NORMALLY sized to operate in the range they are designed to reproduce, i.e. from 13 to 18 cm for the midrange GG and about 2.5 cm for tweeters.
Tweeters with SOFT domes.

These tweeters, which use SILK domes impregnated with damping compounds, came into use at the very beginning of the 70s in the form of the 2.5 centimeter tweeter from Peerless (remember the tweeters of the first Polk Audio speakers!), followed by a similarly sized magnificent tweeter from Audax, which was used in a huge number of British and American developments of the 70s and early 80s.

With the advent of modern titanium and aluminum domes, such designs fell out of favor, which led to Audax's soft-dome tweeters being swept off the market.

In recent years, soft-dome tweeters, oddly enough, have “returned” in the form of the constantly improving Scan-Speak D2905 (2.5cm) series, which, all other things being equal, can compete with ANY metal ones. These tweeters combine vented pole pieces with "advanced" labyrinth loads, new profiles and dome coatings. As a result, they have the sonic resolution and detail of the BEST metal domes, but WITHOUT the characteristic resonance in the 22-27 kHz region.

Pros: Significant self-damping and potentially completely flat frequency response, as well as first-class impulse response. Potentially NATURAL, OPEN sound WITHOUT annoying and tiring resonances, high quality reproduction of digital recordings.

Cons: The first generation of soft dome tweeters sounded DIM and were characterized by an inexplicable FATIGUE listening experience. In addition, the power characteristics were extremely low, which required the use of crossovers with high slope (18 dB/octave). Modern soft-dome tweeters have solved this problem, and the best examples can compare in sound reproduction quality with ANY other technologies, including electrostatic and ribbon tweeters.

The best examples: The Scan-Speak D2905 family of 2.5 centimeter tweeters.

Metal dome tweeters.

Advances in the German metallurgical industry (at Elac and MB) in the mid-80s made it possible to produce thin-profile domes made of titanium and aluminum, and the tweeters themselves began to be produced in Germany, Norway and France. These heads can sound VERY CLEAR, challenging the best examples of electrostats (if done correctly, of course).

The disadvantage of such tweeters is insufficient SELF-damping, and aluminum tweeters in this regard (in the ultrasonic range) are somewhat better than titanium ones. Currently, ALL metal dome tweeters have noticeable peaks in the ultrasonic region, ranging in amplitude from 3 (excellent) to 12 dB (mediocre).

There is much debate about the significance of these ultrasonic resonances, as the engineers at Philips and Sony took great care to ensure that our wonderful "Perfect Sound Forever" recordings did NOT carry ANY music above 20kHz. Without disputing the limitations of the signal source itself, it should be noted that power amplifiers (and VCDs) are capable of generating false ultrasonic signals, especially near or after clipping. These ultrasonic signals can excite the resonance of the metal dome, causing intermodulation distortion that falls into the AUDIOUS range.

Pros: Smooth piston movement down to resonance frequency, resulting in VERY HIGH resolution, TRANSPARENCY and immediacy of sound (if designed well). Dispersion is generally GREAT because metal domes have flatter profiles than soft ones.

Cons: Potentially "metallic" coloration caused by ultrasonic peak intermodulating with audible frequency signals. Some early examples were characterized by limited power capabilities - when overloaded, the break distortion sounds exceptionally disgusting.

Best examples: Vifa D25AG-35-06, a 2.5 cm aluminum dome tweeter that sounds even better if you remove the plastic baffle. Its dome has a ventilated pole piece, and therefore the power response is very decent, and the ultrasonic peak is only 3dB even with the baffle removed (recommended). Focal tweeters are said to be even better.
Tape tweeters.

The rare Kelly Ribbons from the 50s are best known as true ribbon tweeters, but others are also found. These tweeters are the ONLY dynamic drivers with the extremely low mass, extremely high bias uniformity and low distortion found in electrostats. True ribbon tweeters are in a category of their own because the compromises that go into designing conventional tweeters simply don't apply. True, this does not mean that they do not have their shortcomings - there are no free lunches in audio.

The biggest disadvantage of tape tweeters is the SINGLE-TURN “voice coil”, freely suspended in the magnetic gap surrounding it on the sides. This means that the impedance and sensitivity of such a radiator approaches zero (unless, of course, a transformer is used). However, even with a suitable transformer, the sensitivity of the ribbon driver is very low, which was the reason why Kelly added a short horn to the tweeter. Unfortunately, the short horn kills what is best about a ribbon driver, namely ACCURATE impulse response and a complete LACK of resonances.

By combining rare earth magnets with an integral transformer, Raven raised the sensitivity of their ribbon tweeters to a fantastic 95dB/m, which is 10 times the sensitivity of traditional ribbon drivers (and that's WITHOUT using horns!). The waterfall plots produced by MLSSA are also quite impressive, as expected. Distortion is also extremely low: Raven claims less than 1% at 105dB SPL, which is quite impressive.

The ONLY disadvantage of Raven drivers is the need for a high slope crossover. This is a potentially serious problem, since the 4th order (acoustic) slope is already on the verge of audibility and is accompanied by a 360-degree phase rotation in the transition zone. The most radical method of “combat” this is to increase the transition frequency and thus use a wideband midrange driver.

Soft dome midrange speakers.

Well, these are just enlarged (up to 5-8cm) versions of soft-dome tweeters, with exactly the same design, but equipped with a half-barrel frame, which acts as a combined frame and centering washer. Unfortunately, what is good for a tweeter is not entirely applicable to a product enlarged to the size of a midrange driver. In a tweeter, displacements rarely reach values ​​​​of more than 0.5 mm (which is already too much), but the requirements for the third derivative of the displacement (sharp jerk) are very severe, since the tweeter is responsible for the very top part of the spectrum and, from time to time, is subject to ultrasonic clicks, appearing when clipping an amplifier or playing vinyl, or when high-frequency noise and distortion appear from DACs.

In contrast, the midrange (or midwoofer) experiences a much greater need for displacement and acceleration for two reasons: when the frequency is halved, the displacement quadruples, and the energy of the MUSIC spectrum is concentrated in the LOWER-MIDDLE frequencies. Both of these factors in combination lead to the fact that the midrange driver must withstand incomparably more power than the tweeter. And this imposes severe demands on the RIGIDITY of the diaphragm and exposes a simple suspension to vibrating modes.

The reason why conventional cones are provided with a separate frame and an INTERNAL centering washer is to FORCE the cone to reciprocate in PISTON mode. Only VERY EXPENSIVE dome-based midrange drivers designed for professional studio monitors (like ATS) use SEPARATE centering washers. As a result, MOST CONSUMER grade speaker domes have serious PROBLEMS with lateral vibration and other erratic movements. In addition, the silk-doped diaphragm is easily deformed by high loads during acceleration. In the end, no one in their life has ever thought of making bass speakers from silk with additives.

As a result of all these problems, soft-dome midranges measure well, but sound much WORSE than their measurements suggest. Even if we focus only on measurements and ignore all of the above, there are heads with limited bandwidth, which, due to a linear offset not exceeding 2 mm, require crossovers with a slope of 12 dB/octave NOT lower than 500 Hz (preferably 800 Hz). One would imagine that one could get by with a large tweeter that would work well at high frequencies, but ALL soft dome tweeters start to "fall off" starting at 4-5 kHz, which is obviously NOT better than using good modern midrange speakers.

Of course, there are exceptions. For example, there are cone-dome hybrids, such as the 13 cm Scan-Speak 13M/8636 and 13M/8640, as well as the similar Dynaudio 15W-75. These new drivers are actually designed as high quality miniature cones, NOT midrange domes. The only thing they have in common with traditional soft domes is the LARGE dust cap, which actually acts as a high-frequency dome driver.

These new cone-domes feature much higher offsets, much less distortion and a much wider frequency response than the old-style soft-dome midranges. Cone-dome drivers are able to reproduce realistic and transparent sound because they are not made of just one material, but use Kevlar, paper and polypropylene.

Another “special case” is the English professional 7.5 cm ATS domes with a built-in short horn. These drivers use a double centering washer, which eliminates the lateral wobble problem that plagues most soft-dome midrange drivers, as well as greatly reducing intermodulation distortion. Ron Nelson (of Nelson-Reed) recommended these heads as being virtually the best available. But these are VERY expensive heads (about $300 apiece). In addition, they must be manually selected to match the resonant frequencies for the left and right channels.

Pros: NONE. METAL dome midrange drivers have some potential, but they require a sharp cut in the crossovers at both ends, as well as an additional HF notch filter to remove the first (and worst) HF break mode. Note: This does NOT apply to cone dome hybrids or professional ATC heads.

Cons: High distortion, fatiguing sound, high crossover frequencies, limited bandwidth, limited power capabilities, and misleading frequency response measurements. To obtain complete information on these heads, detailed sweep-tone intermodulation measurements and laser holography (vibrometry) are required. Note: Does not apply to cone-dome hybrids or professional ATS heads.
The best examples: Professional 7.5 cm domes ATS are a fundamentally different “beast” than all conventional soft-dome midrange speakers. They cost about four times more (what did you want?). Also, the Scan-Speak 8636 and 8640 are excellent full-range midrangers.

Paper cone midwoofers and widebands.

This variety of heads traces its roots back to Rice & Kellogg's original 1925 dynamic loudspeaker patent. Paper cone heads range in quality from TERRIBLE to GREAT, from 10-cent wheezes glued to motherboard computer up to 13 cm super-quality cone-dome heads from Scan-Speak, classic Lowther, horn-loaded and 30-38 cm Tannoy Dual Concentric.

The oldest of all materials is actually a composite, which significantly changes its properties depending on the processing of a particular material (which is kept secret by each manufacturer). Cone treatment is quite important because UNtreated paper undergoes significant transformations depending on humidity. Additives stabilize the material, improve self-damping and expand the range of action of the cone towards HF.

Pros: Good to excellent SELF-damping, potentially excellent RESOLUTION and DETAIL, potentially VERY FLAT FREQUENCY RESPONSE, and GRADUAL onset of resonant cone break. Can be used PROBLEM-FREE with linear-phase crossovers with low steepness. Paper is a material that sounds BETTER than measurements indicate. And this is a big plus, not a minus.

Cons: The material is not as rigid as Kevlar, carbon fiber or metals, so the heads may lack a little of the super detail found in electrostats. It is characterized by lower “loudness” than the above materials, but it enters the resonant break much smoother and more gradually. For best results, paper cone drivers may require moderate shelving EQ in the crossover.

Paper is not as dense as synthetics, and therefore pair matching is not always strict, which to some extent can affect the formation of the scene (depending on the accuracy and quality of workmanship). The properties of the paper can change slowly over time (depending on the composition of the composite that covers the cone).

Best samples:

13 cm cone-dome midrange Scan-Speak 8640 with linear frequency response from 300 Hz to 13 kHz, very low distortion, excellent impulse response and detail.

16.5 cm high sensitivity Audax PR170M0 (100dB/m!!)

16.5 centimeter Diatone PM610A (anniversary edition) from Mitsubishi. This is an ultra-wideband head covering the range from 70Hz to 20kHz with the most common acoustic design.

Various Lowther models. They require a load on the horn to operate properly, as well as to avoid damage when moved. If done correctly, they can cover the range from 50Hz to 18kHz.

Midbass from Bextren.

Bextren is an acetate plastic derived from wood pulp, which is usually damped with a layer of additive coating on the front of the cone to keep the first strong resonance of the diaphragm in check at 1.5 kHz. The material was originally developed by the BBC in 1967, which (as a more dense and “predictable” material) was supposed to replace paper in monitor acoustics. The material became widespread in the early 70s. Typical speakers of that time consisted of a 20 cm Bextren midwoofer from KEF or Audax and a 2.5 cm soft dome tweeter from Audax.

In BBC developments, notch filtering has ALWAYS been used in order to achieve a smooth frequency response of Bextren heads in the midrange. The most famous (or infamous, depending on whether you are a listener or a developer) is the head from KEF B110, installed in BBC LS3/5a minimonitors. Few people know that these speakers, which are famous for their “sweet midrange,” use a notch filter that provides an attenuation of 6 dB at a frequency of 1.5 kHz.

Over time, Bextren was replaced by the polypropylene developed by the BBC, which is characterized by a much smoother frequency response, does NOT require additives (in the coating) and, thanks to the reduction in the mass of the cone, provides a 3-4 dB increase in sensitivity. Today Bextren is considered an obsolete material by almost ALL speaker manufacturers.

Pros: Good repeatability, potentially excellent "scene" (by mid-70s standards). The resolution is higher than many paper cones.

Disadvantages: very low sensitivity (82-84 dB/m), the need for notch filtering in the midrange region, “quacky” sound color by modern standards, sudden, extremely unpleasant appearance of resonant breaks even at a moderate volume level, as well as a large number of resonances in the midrange region HF

Best examples: NONE. Modern designers are unwilling to put up with the low sensitivity, complex notch filtering and shelving equalization that is necessary to make the heads sound at least acceptable. And although traditionalists praise the KEF B110 heads used in the Rogers LS 3/5a, the terribly uneven frequency response of these heads requires the use of crossovers of extraordinary complexity. In any case, modern Vifa P13WH-00-08 heads are a hundred times better than any B110.

Midwoofers made of Polypropylene.

This material was developed and patented by the BBC in 1978 as a replacement for Bextrain. Thanks to the “innate” SELF-damping, a well-designed polypropylene head is able to provide a flat frequency response in a given range without (or with a minimum) equalization at all. In addition, polypropylene heads typically have sensitivities ranging from 87 to 90 dB/m, which is a significant improvement over Bextren.

Because it requires a MINIMUM amount of “hands-on,” this material has become almost universally used in column construction. The only challenge along the way was creating a cyanoacrylate adhesive that would “stick” to such a slimy material as polypropylene. But this problem was resolved in the early 80s.

As with paper, this cone material is used in speakers ranging in quality from mainstream boomboxes to high-end speakers like the ProAC Response line. The profile of the cone, the “cut” at the edge of the cone and additional materials-additives to polypropylene are the determining factors in the quality of the heads.

Pros: Very smooth frequency response (if everything is done correctly), very low sound distortion, good impulse response, gradual onset of the resonant break of the cone, high sensitivity, as well as simplicity of crossovers, which can consist of a single capacitor for the tweeter. The best examples can be as clear as the best paper cones, which is a very high standard.

Cons: In terms of transparency, they are somewhat inferior to the class of heads made of hard materials and planar electrostatics. Many polypropylene midwoofers do not “get along” well with popular metal dome tweeters - a large difference in resolution can be striking to the ears of a trained listener. Polypropylene is not the best choice for woofer with a diameter of 20cm or more (unless, of course, polypropylene is reinforced with another, more rigid material). Woofers measuring 25cm or larger are best made from stiff paper or carbon fiber.

Best samples:
18 cm Scan-Speak 18W/8543 midwoofer (the same as used in ProAC Response 3.x). Apparently this is the BEST polypropylene head in the world.

Another contender is the 18-centimeter midwoofer DynAudio 17W-75, used in the Hales System Two Signature.

The Vifa P13WH-00-08 mid-bass/midrange driver is also an excellent “performer”, well suited for use in minimonitors. Its uniqueness lies in the fact that it has a completely flat frequency response in the midrange region, accompanied by a VERY SMOOTH 2nd order Bessel cut. This Vifa does NOT have that characteristic "plastic" sound - it sounds more like a fine paper cone.
Midwoofers made of hard materials.

Heads made of Aluminum and Magnesium.

The first heads that were used to a limited extent in hi-fi were small 5 cm aluminum cones by Jordan Watts. Manual assembly, high price and low sensitivity were highly limiting factors preventing their widespread adoption in the market.

There is a new generation of British and German two-way minimonitors that use 13-16.5 cm midrange drivers with aluminum cones. These heads usually have VERY LOW sensitivity (82-84dB/m) and almost ALWAYS have high-Q resonance in the upper part of the operating range. These heads are NOT sold separately on the open market.

IN new series Seas Excel heads use magnesium cones with an “intriguing bullet” of solid copper instead of the usual dust cap. Without a doubt, it all looks cool and, unlike the aforementioned aluminum diaphragms, it is also characterized by a good sensitivity of 87 dB/m. True, at 4.9 kHz there is a strong resonance (16 dB in value) - so you will have to tinker with the crossover.

Carbon fiber heads.

These very sensitive and very expensive (approximately $300 each in 1980) hard cone drivers were made famous by Japanese manufacturers, who first used them in TAD PROFESSIONAL STUDIO monitors with 30.5 centimeter diameter woofers. Today, carbon fiber prices have fallen, and Vifa, Audax and Scan-Speak are producing very good examples. The Japanese were the pioneers of this technology and produced a huge number of heads of this type. However, if you are NOT a Japanese manufacturer, it is almost impossible to get them.

These drivers are characterized by TRUE piston action and absolutely OUTSTANDING bass and midbass (you will NOT hear better ANYWHERE!), but, of course, are characterized by the characteristic double resonance at the edge of the working area. Unfortunately, these peaks are very audible on most carbon fiber heads and, even worse, cannot be eliminated by a notch filter tuned to a frequency that lies BETWEEN the two peaks, requiring TWO notch filters to “curb” the peaks and make them inaudible. filter or crossover with a VERY large cutoff slope (24 dB/octave).

And although working with drivers that require such complex crossovers (which can contain up to 50-60 components) is extremely difficult, it is worth recognizing that carbon fiber woofers are the ONLY drivers that produce truly TOUCHABLE bass.

The Scan-Speak 18W/8545 heads look quite interesting: although they clearly have the “birthmark” of all carbon fiber heads - the double peak - they look quite well damped, and the fracture area ABOVE these two peaks looks very smooth. Perhaps these Scanspeaks can be used even with a conventional second-order filter.
Kevlar heads.

Kevlar heads appeared in the mid-80s in the French Focal and German Eton lines, the latter featuring superdamping thanks to a high-loss Nomex honeycomb structure separating the front and rear layers of Kevlar.

Currently, 18 and 20 cm Scan-Speak have the best frequency response and the lowest intermodulation distortion of ALL Kevlar heads. Another nice feature of this family of heads is the well-controlled cutoff AFTER the Kevlar resonance frequency. ALL OTHER Kevlar heads, except Scanspeaks, are characterized by significant chaos in this area, which, without a doubt, does not have the best effect on the smoothness of the frequency response and sound transparency.

Heads made of composite materials.

Audax has an unusual technology called HD-A. The base is an acrylic gel containing a mixture of lined (the fibers are orderedly “combed” and laid out) carbon fiber and Kevlar. The frequency response of such a system is impressive in its evenness. The only exception is a moderate peak at the end of the range.

Another interesting series of composite heads is Focal Polyglass, in which the cones are made from a mixture of paper and carbon fiber. The most striking example is the 6V415 midwoofer with an exceptionally flat frequency response and a fairly good offset. For fans of triodes who do not recognize crossovers and dream of wideband heads, a bunch of four 4V211 with a band from 60Hz to 12-14kHz may be perfect.

Strong and weak sides hard heads.

Pros: THE BEST transparency, stage and depth that can be matched (or even surpassed) only by electrostats. The best examples have high sensitivity, high peak levels and very low intermodulation distortion. This class of heads is considered by many designers to be unsurpassed, and excellence should continue to increase as technology develops.

Cons: Older samples were characterized by strong peaks at the edge of the operating range, and most also had uncontrolled chaotic bending behavior ABOVE the resonant bend, which led to rapid sound fatigue, as well as a decrease in depth and lack of “air”.

Speakers that do NOT use well-designed notch filters using metal, Kevlar or carbon fiber heads can and should be considered illiterately designed, since the narrow high-frequency peak cannot be corrected by standard high-pass filtering. The audibility of such a peak is obvious to ANY listener who is familiar with the sound of UN-equalized Kevlar or carbon fiber heads (just tap the head to get an idea). The new 18cm Scan-Speak 8545 or 8546 heads are perhaps the first of a new generation of heads that DO NOT require a notch filter.

Although these types of heads play quite loudly, the onset of the resonant break of the diaphragm can be VERY sharp and unpleasant, akin to amplifier clipping. Some Kevlar or carbon fiber heads require extremely long break-in times (over 100 hours), i.e. to soften the fibers in the cone and centering washer.

Top picks: The Scan-Speak family of 13, 18 and 22cm Kevlar heads, plus the new 18cm carbon fiber-paper composite heads. These heads are the ONLY heads with well-damped peaks and good behavior at the cutoff AFTER the main HF peak. In addition, Scanspik heads also have screw-in copper-plated pole pieces, which leads to a reduction in inductive types of intermodulation distortion by TENS of times or more.

The new Seas Excel series of heads with magnesium cones and an all-copper phase plug (instead of a dust cap) is also very interesting, if, of course, the designer wants to seriously tinker with the development of a very deep and PRECISION notch filter to correct the first resonance of the magnesium cone.

How to choose a head.

It is best to use a method that is so crude that you can’t imagine it being more effective. You need to mount the head under study on a flat panel with the recommended IEC dimensions: 135 x 85 cm and just listen to it. WITHOUT crossovers, WITHOUT acoustic design, and if we are talking about a tweeter, then there is no need to make it loud. And you need to listen to PINK NOISE (which will help assess the visibility of the peaks visible on the frequency response and frequency response) and just music (to assess the “musicality” and “resolution”).

This will require trained ears, though, as you'll have to listen out for peaks that might be removed by a crossover without being hampered by the limited bandwidth of a separate head. However, this listening process will tell you a lot about how complex

It is also necessary to carry out a number of measurements (using MLSSA), namely to remove:

1) Impulse response. How quickly does the head “stop”? Is there a chaotic mess in the attenuation zone, or is there just one smooth resonance there? Or maybe there are two or more resonances?
2) Group delay time and frequency response. How distorted is the frequency response AFTER the first resonant break? Can this be fixed with a crossover?
3) Frequency-transition response (“waterfall”). Are you willing to accept resonances that can NOT be suppressed by a crossover? If crossover correction is necessary, how complex should it be?
4) Amplitude-frequency response. Take a close look at the Fletcher-Munson curves (FMC) - they clearly demonstrate the area where hearing sensitivity is highest and responds to even the slightest changes in sound pressure (unevenness in frequency response). The MOST CRITICAL region is the band from 1 to 5 kHz - ANY PEAK in this region (even half a decibel!) will be AUDIOUS and will be perceived as an unpleasant “canned” quality. In contrast, small dips are much less noticeable to the ear (unless, of course, they are caused by reflections or resonances).

It is in attention to such “little things” that the BIG DIFFERENCE between an AMATEUR HOMEMAKER and a SERIOUS DEVELOPER SPECIALIST lies.

If we return to the very complex field of crossover design, then developers belonging to the OBJECTIVE school of development usually prefer 4th order Linkwitz-Rieley circuits, which provide the smoothest and most accurate frequency response shape, as well as the best suppression of out-of-band intermodulation (due to distortion of the pulse shape and slips)

Special thanks to Laurie Fincham of KEF for her research into precise computer electroacoustic modeling of driver behavior combined with crossovers, which allowed us to synthesize and optimize 2nd, 3rd and 4th order acoustic cuts. Before Fincham, crossover development was essentially just “tailoring” STANDARD crossovers from teaching aids under real conditions, which gave only a rough approximation. After Laurie Fincham published his technology, choosing a crossover topology and the desired "target slope" of the cut became a matter of simply choosing from several options, after which the trial and error work was left to the computer, and you just had to make a final decision on the results.

Of course, back then, in the early 70s, when this path was just being blazed, a “computer” meant a specialized mini-complex from Hewlett-Packard costing $150,000, as well as a Fortran programmer on staff who punched holes in card decks and generally managed it all farming Nowadays, the technology for optimizing crossovers has become much simpler and cheaper - take any decent computer and load it with one of the programs like XOPT, CALSOD, LAEP, etc. As a result of such a dramatic reduction in capital investment in both crossover optimization and powerful measurement systems, it is a shame for modern speaker builders to complain about life, regardless of the philosophy underlying a particular speaker production.

Developers belonging to the objective “school”, until recently, completely neglected impulse response, diffraction controllability, as well as all those murky areas like special capacitors, special inductors and special wiring cables. On the contrary, research is focused on constantly improving the quality of heads, reducing cabinet resonances and the highest possible PAIR precision in production.

“School” of impulse shape preservation.

This school includes manufacturers such as Dunlavy, Thiele, Spica and Vandersteen. Developers pay close attention to controlling diffraction, spatial displacement of the heads for the purpose of simultaneous arrival of signal components to the listener, and also, as a rule, use 1st order crossovers (with a cutoff slope of 6 dB/octave). True, some, like Spica, can also use Gaussian or Bessel crossovers of the 3rd (18 dB/octave) or 4th (24 dB/octave) orders.

Acoustics ONLY of this design are capable of accurately maintaining the shape of the pulse, sometimes surpassing electrostatic or tape emitters in this matter. However, the issue of AUDIOUS phase distortion (pulse shape distortion) is EXTREMELY CONTROVERSIAL even in the AES environment, despite the fact that MANY engineers consider solving this “problem” a WASTE of time and money.

In a typical linear-phase (preserving the pulse shape) speaker drivers MUST have excellent characteristics, including in a range that is 2 octaves beyond the nominal frequency range, so whatever one may say, both intermodulation distortion and power are sacrificed characteristics. In order to at least partially solve this problem, very EXPENSIVE heads are required, as well as extremely accurate “addressing” of resonances in crossovers. Controlling the directivity of the radiation when using first-order crossovers and moving the heads in space is also a task fraught with difficulties. As a result, this type of speaker can sound SIGNIFICANTLY DIFFERENT depending on whether you are sitting or standing, and also to the side of the main axis.

The most serious MISTAKE when designing such speakers is the “immersion” of midrange drivers or tweeters into cavities covered with felt in order to shift the heads in depth within the framework of TRADITIONAL acoustic design. Felt “works” very well as a damper INSIDE the case, but expecting 100% absorption of a broadband signal from it is sheer stupidity. NONE of the known absorbers is characterized by 100% absorption over the entire spectrum - the best that can be achieved is attenuation of 20-30 dB, and then only in a certain frequency band. In addition, such weakening is achieved by a COMBINATION of several materials, with a total THICKNESS of TENS of centimeters! So imagine how small the effect of a 1 centimeter thick strip of felt is.

Placing the head deep into a cavity with a HARD surface will give a very clearly audible color sound in the form of “hunching”, similar to the color that your voice acquires when you fold your hands into a mouthpiece and yell. Lining such a cavity with a thick layer of felt, of course, helps a little, but the “gundos” is still present if you listen. But that's not all. Placing the felt in close proximity to the diaphragm LOADS it with additional “mass”, reduces sensitivity and worsens transient response. And all these “antics and jumps” in order to observe something similar to a square wave on an oscilloscope???!! Nonsense! If you really need to move heads, do yourself a favor and use SEPARATE boxes and panels for this (for midrange drivers and tweeters)!

With the right approach, linear-phase systems can sound as “open” and “free” as electrostats. Especially if the design is characterized by low diffraction. A disadvantage may be the limited dynamic range of the tweeter and midrange driver, as well as a complex polar pattern, which results in a rather narrow zone of optimal listening position.

Horns and other highly sensitive systems.

As already mentioned, the speakers of the mid-50s were very, very sensitive by modern standards. The favorites of modern audiophiles - electrostats, planars and minimonitors - are all characterized by sensitivities of about 82 dB/W/m (i.e., they have an efficiency of about 0.1%). While the most popular speakers of the 50s had a sensitivity of about 92-96 dB/W/m (i.e., an efficiency of about 2%). The largest and most prestigious systems that could still be used at home were characterized by a sensitivity of 102 dB/W/m (10% efficiency).

So what happened? In the 50s there was a strong belief that the best speakers were the most sensitive. Audiophiles of the time were well aware that the best speakers were produced by companies such as Western Electric, Altec and RCA. If anyone needed proof, they could go to the cinema and experience Cinerama, Ben-Hur or 20,000 Leagues Under the Sea.

The belief in sensitivity as the highest virtue was shaken by the appearance on the market of the AR-1, the first SMALL speaker in the acoustic suspension type, which was very competently designed. And although it was 10 times less sensitive than most of the then popular speakers, it REALLY played up to 30Hz and did not “thump”. And at the same time it was compact!!! True, we had to wait a little while until amplification technology began to cope with the “requirements” of speakers with “acoustic suspension” - by the time STEREO as such appeared in the late 50s, amplifiers with a power of 60 W per channel began to arrive on the market. This has led speaker designers to sacrifice sensitivity in exchange for better damping and handling. Now that the long-awaited goal of speakers characterized by NORMAL measurements was within reach, many developers began to look for better damped, but less sensitive materials.

That's when the famous "East Coast Sound" vs. "West Coast Sound" epic began. “Westerners” were represented by JBL, Altec and Cerwin-Vega, and “Easterners” were represented by AR, KLH and Advent. In the 60s, Westerners produced smaller and smaller speakers that didn’t care about the sensitivity and dynamics of high-quality theater systems, but deliberately copied and exaggerated “last year’s” droning bass and horn coloration of the giants. The most glaring example of this marketing philosophy is the highly successful JBL L100 speakers - wonderful looking bookshelf speakers with a bright orange textured foam grille. And everything was fine until... they were turned on.

The Eastern school of speaker design, in response to the “harshness” of the sound of the first transistor amplifiers and the aggressiveness of Westerners, gradually made their speakers sound increasingly dull. Just like the creations of the Westerners, the speakers of the Easterners were now characterized by smooth frequency response, but, unlike the Westerners, the Easterners did not pay any attention to distortion, improving crossovers and strengthening the cabinets. If you open one of these speakers, you will see a cheap electrolytic capacitor on the tweeter, a piece of cotton wool or glass wool, a complete absence of bracing (internal bracing of the case), as well as thick grills stretched over massive overhanging subframes. In the early 70s, American audiophiles were very tired of the ill-conceived developments and mediocre build quality of both Westerners and Easterners, and began to look for “happiness” on the other side of the Atlantic. For example, in Britain...

In the 1970s, England was at the forefront of global speaker design, researching new materials (such as Bextrain), using computer simulations of crossovers, and using FFT (fast Fourier transform) to track resonances in loudspeaker heads and cabinets. By that time, amplifiers with a power of 120 W had appeared, which gave manufacturers the opportunity to “throw off” a couple more dB from sensitivity, but thereby reduce resonances and smooth out the frequency response.

The lower limit of sensitivity was reached in the early 70's with the BBC LS 3/5A, which were very SMOOTH and ACCURATE sounding speakers that were in great demand by audiophiles around the world.

However, not everything was so smooth... The heavily damped 13-centimeter Bextren cone used in the B110 “increased” the sensitivity of this little miracle to the lowest 82dB/W/m. In order to “open up” and “play”, such a speaker, no joke, required an amplifier with a power of 200 W. Many a happy LS 3/5A owner had amps that were healthier and heavier than the speakers! And yet, THEN it was a VERY good choice in terms of speakers. And amplifiers got better and better over time, so the problem of choosing between a piercingly bright or dull and dull sound simply disappeared.

A couple of years later the KEF 104A appeared. KEF modified the midrange and high-frequency drivers from the LS 3/5A, developed a new bass driver as a continuation of the B139 and released a speaker in this configuration, in which a computer-optimized Linkwitz-Riley crossover was installed for the first time. Although the sensitivity was still no higher than the LS 3/5A, the 104A set new standards for natural sound, clarity, and stereo imaging (which was a direct result of the "advanced" crossover).

The trend towards increasing speaker sensitivity and reducing power input began in the late 70s with the advent of more “efficient” cone materials such as polypropylene. In the late 80s, the tube renaissance began, manufacturers had a wide choice of materials for the manufacture of speaker heads, including the newly “discovered” paper with additives, the already mentioned polypropylene, Kevlar and carbon fiber.

The new materials did not require any external damping with all sorts of compounds, such as, for example, products from the same Bextren, but completely “relied” on internal SELF-damping (based on the properties of the material of the cone itself). Laser vibrometry and computer modeling ultimately led to a number of noticeable improvements in materials for cones, tweeter diaphragms, and also significantly improved the methodology for designing gaps in magnetic systems of heads with ventilated pole pieces. Currently, the best direct radiation heads are made by Scan-Speak, DynAudio, Audax and Focal, which are characterized by sensitivity of about 89-94dB, which is four times more effective than in the 70s.

Joe Roberts' Sound Practices magazine had a great influence on the North American market by introducing the developments of Japanese, Italian and French schools to it. Overseas audiophiles had NO “memories” of the marketing disaster of “West Coast Sound” and therefore continued to have a strong respect for classic high-sensitivity theater loudspeakers.

If you leave the Anglo-American orbit, you will notice that developments in the style of the “old” theater speakers Western Electric, Altec and JBL studio monitors, as well as P.G.A.H. Voight is still taken quite seriously. And their attractiveness is not explained by nostalgia. The latest drivers and horns made from exotic materials are hitting the market at prices that would astound Western audiophiles. These “alternative” speakers work especially well with dead amplifiers that use single-ended, directly heated triodes (a 3-watt amplifier will simply “die” on electrostats or half-room-sized planars, but will “open up” perfectly on a full horn system with sensitivity 104dB).

For those who believe that amplifiers have already reached the ceiling of perfection (and that is almost all AES members and home theater manufacturers), all this “throwing into the arms” of archaic “alien” technologies looks like some kind of fad or joke. Sugary murzilkas explain the “phenomenon” of horns/triodes by the fashion for retro, which is another example of the mythologization of the past that is fashionable today.

The other side of the coin is that MOST sane adherents of horns with triodes already had and then abandoned REGULAR, standard audiophile systems. As a designer of fairly standard speakers, I can personally attest that increasing the sensitivity of conventional drivers is definitely worthwhile - you will get a significant increase in clarity, spontaneity and naturalism of sound, and the choice of an amplifier will be greatly simplified, and in addition, there will be opportunities to try some very interesting technologies.

From a technical standpoint, horn-loaded drivers typically have very low harmonic and intermodulation distortion, flat frequency response, time-domain reflections, and very steep cutoff characteristics at both ends of the operating range. From the perspective of mainstream high end designers, horns suffer from serious problems in terms of impulse response, diffraction and uniform dispersion.

The root of all these problems, especially with CHEAP PA horns, is the acoustic reflection from the edges of the horn mouth. When a sound wave hits a sharp edge, it diffracts (bends around it) and is re-emitted in all directions (as if there was another head at the reflection point). The wave reflected from the mouth of the horn gets back deep into the horn, where, as a rule, there is a phase plug or a head with a rigid cone, from which it is again reflected and goes out. This sequence of reflections is called stepwise and is audible much better than what can be expected from small irregularities in the frequency response.

Although the frequency response does not fully show the audibility of step reflections, they are clearly visible on the impulse response or on the frequency-transition (“waterfall”) response. Everything can be seen especially well if you measure the horn directly, without a crossover. Inexpensive PA horns that are too short suffer from this affliction the most, and as a result have the roughest "horn color" of all.

But there is a solution for this problem too. If you are willing to sacrifice 1-2dB of sensitivity, you can line the inside of the horn with thin felt (3 millimeters thick). The deeper you can push the felt, the better the damping will be. True, if you overdo it, then the lower part of the horn's operating range will sag, and with it the sensitivity. It will be very good if you can control the “improvement” process using MLSSA - just keep an eye on the impulse response.

The best solution, of course, is to completely eliminate the reflection from the mouth, which is achieved by giving the horn a tractrix shape, which was invented by Voight back in the late 20s!!!

The tractress also has a sharp edge at the mouth, only at the moment when the sound is reflected from the boundary, the horn is already bent by 90 degrees, as a result of which it is almost impossible for the reflected sound to “get” to the phase plug and be reflected again. Hence the complete absence of standing waves, only moderate reflections and virtually no “horn coloration”, unless of course the compression driver (what is loaded onto the horn itself) is properly designed.

Attention: there are a lot of horns on the market with a tractrix profile, BUT this profile is provided only in ONE direction (for example, vertically), and therefore at the other two edges of the mouth the problem remains unresolved, which negates the entire advantage of the tractrix. Creating a horn with a square or even more so a round mouth and a tractrix profile is another problem, the solution of which costs an enormous amount of money.

For example, the BEST 5cm drivers for horns operating from 500Hz to 22kHz are made by JBL and TAD for use in studio monitors, and they cost EVERY 800 bucks, NOT including the horn itself!!! Compare that to the price of a top-of-the-line head from Scan-Speak at 120 greens and feel the difference. No, of course, you can buy an entry-level PA horn for 80 bucks, but for that money you will get strictly appropriate quality. Don't even think about getting sound from a PA horn comparable in quality to that of a JBL studio monitor head with a 5cm titanium diaphragm. They may be similar EXTERNALLY, but INSIDE they are completely different.

The thing with horns is that the difference between “bad” and “best” is REALLY GREAT, and much more obvious than the difference between standard heads. Moreover, the best horns are steeply expensive, require machined Alnico magnets, diaphragms made from exotic metals, and the horns themselves have complex shapes machined with extreme precision. There is no need to talk about cost reduction here. And near the bottom "bar" we have low-grade PA horns that require serious modification and improvement work before the term "hi-fi" can even be applied to them.

Despite all the mainstream trends, I think the market for horn drivers will continue to grow over time. Most likely, it will be driven by companies like JBL, Altec and Tannoy, which can afford to spend a million or two dollars on machining and start-up costs.

Minimalists.

This group includes some Italian, Scandinavian, English and American speaker manufacturers. In such speakers, crossovers are extremely simple, sometimes reduced to one single capacity! The heads and components of the crossovers are characterized by the highest quality, as well as exotic materials are used for the manufacture of cabling wires and housings.

When creating speakers of this type, measurements, as a rule, are not used at all. Since this design philosophy leaves head resonances without any correction at all and, thanks to minimalist crossovers, allows for the presence of all sorts of “jambs” in the final frequency response and impulse response, the final sound will most likely depend heavily on the sound signature of all other parts of the audio system .

And while not many developers are TOTAL minimalists, the idea that “parts quality matters” has INFECTED just about everyone else in the industry. Nowadays, it seems, not a single “self-respecting” high-end manufacturer DOES NOT use electrolytic capacitors in crossovers, and mylar capacitors are rare. This is a significant difference from the 70s, when the crossovers of even the MOST technically advanced speakers used components that, by the standards of today’s “traditions,” would seem like garbage - that’s what insanity has reached. 20 years ago ALL attention was focused almost exclusively on HEADS, development technologies and BODY design. Today, when there is nothing more to study in this area, the developers have gone to great lengths, researching the “influence” of everything else on the sound, right down to the SCREWS that secure the heads, as well as the gold plating of the contacts. Horror, and that's all.

Who invented the very first speakers?

The very first speakers were invented by Paul Klipsch. What do you think the world's first speaker systems were?

With dynamic heads, electrostatic or maybe planar?

Neither one, nor the other, nor the third. The history of acoustics began with horn speakers, and this happened when a person first put his folded palms to his lips to make his speech sound louder.

Actually, the hands played the role of a mouthpiece, and the vocal cords played the role of a speaker. As simple as this example is, it really demonstrates how it works...

The first devices for amplifying and reproducing sound used the horn principle. Take, for example, such a device as a megaphone (in those days, of course, without an amplifier), or wind musical instruments.

And the famous gramophone, which became almost the first device for playing music. Are there enough examples? And only then, starting in the 30s of the twentieth century, the first outlines appeared in the design of other types of acoustic systems.

If you ask any person interested in audio equipment about horn speakers, then with a high degree of probability he will immediately remember the Klipsch brand. And he will be absolutely right, because this company has a great many inventions in the field of acoustics that have glorified its name.

The company was founded in the 40s of the last century, just at the time when the first developments in amplification technology appeared. The company was founded by Paul W. Klipsch, who was the main ideologist and developer of the company for many years.

It should also be mentioned that Hemholtz made a great contribution, and it was from him that the theory about sound amplification originated... A number of Helmholtz’s technical inventions bear his name. The Helmholtz coil consists of two coaxial solenoids, separated by a distance of their radius, and serves to create an open, uniform magnetic field.

The Helmholtz resonator is a hollow ball with a narrow hole and is used to analyze acoustic signals, and also in the construction of low-frequency sound speakers to enhance low frequencies or, conversely, to suppress unwanted frequencies in rooms.

The most important design elements of loudspeakers have remained unchanged since their invention at the beginning of the last century. Modern electroacoustics came onto the market with the invention of the telephone by A. G. Bell and T. Watson in 1876. And although since then the improvement of electroacoustic transducers (that is, loudspeakers) has been the topic of an endless series of scientific research and articles, much more than those devoted to any other element of the sound reinforcement circuit, there are practically no fundamental changes.

Speaker history began back in the 19th century. The first patent application for an electrodynamic moving-coil design was filed in 1877, and for an electrodynamic loudspeaker in 1898. However, these inventions did not receive practical application at that time - there was not yet a sufficiently powerful source that would allow the head of a loudspeaker with a moving coil to be driven.

Commercial models did not appear until the 1920s, when tube amplifiers became available. The first electrodynamic loudspeakers had high-impedance coils, fabric suspension, and powered electromagnets. DC. Some historians of technology indicate that the first electrodynamic head as close as possible to its modern design was patented in 1925 by the company General Electric.

Externally, the designs of dynamic drivers for reproducing low and high frequencies differ, but contain the same components. The LF head has a metal (less often plastic) frame, which is also called a basket for its shape or a diffuser holder - this is its purpose. The diffuser holder windows provide free air movement at the rear of the diffuser. In the absence of windows, air could act on the moving system as an additional acoustic load, reducing the output in the low frequency region. The manufacturing technology of the diffuser holder is determined by the power and size of the head. The main requirement is to ensure a rigid structure, free from vibrations that can cause overtones. From this point of view, it is better to use cast structures made of metals or composite materials. A conical diffuser is fixed to the frame, usually made of paper (actually from shredded wood), pure or filled plastic, and less commonly, metal or ceramic. A sleeve (impregnated paper or metal) is attached to the rear (narrower) part of the cone, onto which the voice coil is wound.

The voice coil is usually wound in two (less often - four) layers with copper or aluminum wire in enamel insulation on a frame (sleeve) and secured to it with varnish. Typically, a standard round wire is used, but for very powerful heads, a wire with a rectangular cross-section is used, providing almost 100% filling of the gap. Modern materials are widely used when assembling the moving part of the head. For example, UV-curing polymer adhesives are used to bond the voice coil frame to a ceramic or metal dome cone. The coil terminals are connected to contacts on the connection board using special, very flexible wires.

Despite continuous research in the field of materials science, most LF and MF drivers, which have a similar design but differ in size, use cone diffusers made of paper pulp. In addition, materials such as polypropylene, bextren, and, more recently, light metals (aluminum, titanium, magnesium) are used. Firms with a name and history, having their own research centers or ordering development, are actively experimenting with various fillers and composite materials, creating combined diffusers. Here, the most famous example is the B&W midrange drivers with a diffuser made of woven Kevlar with impregnation. Straight-line cones were used in low-frequency drivers only in the very first drivers. The rigidity of such a design is not enough for the entire operating frequency range, and above a certain frequency the radiation acquires a bending character: only its central part actually works. The diffuser is too heavy and too soft to accurately follow the movement of the coil. It simply does not have time to completely deflect and return, and bending vibrations give rise to overtones and additional coloration of the sound.

The simplest and ancient way To combat this phenomenon, a series of concentric grooves are formed on the surface of the cone during the manufacturing process. Modern loudspeakers use a whole range of measures to suppress parametric resonances. Firstly, almost all diffusers have a curvilinear generatrix. Secondly, more and more of them are made of materials that effectively dampen longitudinal vibrations and, in addition, they have a variable cross-section: the coil has a larger cross-section, and the suspension has a smaller cross-section. Of course, it all depends on the chosen material. For a paper diffuser, a special impregnation is suitable, and for a layered or composite structure, the combination of physical and mechanical properties of its constituent materials is important. Since the range of reproduced frequencies of a loudspeaker head is determined by the area of ​​piston movement of its diffuser, it is important that it be as rigid as possible, but at the same time have a minimum mass.
The external suspension of the diffuser, which ensures its translational movement during operation, can be made as a single unit with the diffuser (in the form of a corrugation with one or several grooves) or as an independent ring made of rubber, rubber, polyurethane and other materials with similar properties, which is then glued to the outer edge of the diffuser. The suspension, especially the low-frequency head, must have great flexibility: this ensures a low self-resonance frequency. Almost immediately below this frequency, the efficiency of the head drops sharply, that is, its own resonance determines the limit of bass reproduction.

The second main requirement for a suspension is that the elastic properties must maintain linearity over the entire range of movement of the moving loudspeaker system.

For quite a long time, high-frequency heads had the same conical diffuser, only smaller. However, today the most common type of HF head is the dome diffuser. It can be soft (made of textiles, such as impregnated silk) or hard - made of metal or ceramics. The design of a typical tweeter differs not only in the size of the cone. Typically, a dome cone with suspension is manufactured as a single unit, to which a sleeve with a voice coil is glued. At the same time, the design does not have a flexible centering washer. The magnetic system, like the diffuser, is fixed to the front flange plate.

Dome diffusers, which can be convex or, less commonly, concave, are made by pressing from natural or synthetic fabrics with mandatory subsequent impregnation. HF diffusers - heads made of synthetic polymer films or metal foil - are becoming increasingly common. To increase rigidity, diffusers are made by vapor deposition of various materials: boron, beryllium, gold and even diamond. There are numerous examples of dome diffusers made from ceramic, which is essentially an oxide of metals such as aluminum.

The centering washer is an indispensable part of the woofer or midrange head; its task is to ensure the correct position of the sleeve with the voice coil in the air gap of the magnetic system. The requirements for the washer are the same as for the suspension - maximum flexibility in the axial direction and preservation of linearity throughout the entire range of movement, complemented by the requirement of maximum rigidity in the radial direction. To increase the efficiency of the head, the gap must be minimal, and the slightest displacement in the radial direction will inevitably lead to jamming of the voice coil. Throughout the process of improving heads, the centering washer was made from different materials(cardboard, paper, textolite, fabric). Today, almost all heads have a centering washer with concentric grooves, pressed from fabric and then impregnated.

The most important element of design and dynamics, which largely determines its electroacoustic characteristics, is the magnetic system. It is formed by a ring magnet located between two annular flanges and a cylindrical core, which forms an air gap with the front flange. The design of a magnetic system with a core magnet, widespread in the middle of the last century, is now practically not used in heads designed for multi-way speaker systems. The magnetic system creates a constant magnetic field in the gap. When a signal is applied to the coil, its magnetic field interacts with the field of the magnetic system, causing it to move back and forth depending on the direction of the current and move the diffuser attached to it. The gap should be as small as possible: this increases the efficiency of interaction between the coil and the permanent magnet.

The magnetic field of a system with a ring magnet is not completely closed in the magnetic circuits. This design has an external leakage field that can affect other devices, such as a color TV picture tube. Therefore, in the case of using such speakers in home theater speaker systems, an additional magnetic screen is required, which is a glass of soft magnetic material that covers the entire magnetic system from the outside.

The shape of the pole pieces (the holes in the upper flange) and the core determines the magnitude of the magnetic induction in the air gap and the uniformity of the magnetic flux distribution in it. The degree of heating of the voice coil and, consequently, its heat resistance depend on the size of the elements of the magnetic system and the width of the air gap. There are conflicting demands here. To improve ventilation, you need to increase the gap, but this reduces the sensitivity of the head and requires a larger magnet. Here there is a field of activity for finding a compromise engineering solution. Therefore, for example, in powerful LF heads the coil diameter is larger, and two ring magnets are often used.

As is known, for efficient work The woofers need to ensure that the sound waves from the front and rear sides of the diffuser are isolated (see “Acoustic Design”, S&V, 4/2004). That's why central hole conical diffuser is closed with a cap, which, due to additional function called dustproof. In some designs, a hole is made in the central core of the magnetic system, closed with a sound absorber, and thick fabric or non-woven material with a large acoustic resistance. Piston movement of the diffuser over a wide frequency band is possible only with its ideal rigidity. For real diffusers, due to the occurrence of longitudinal vibrations of the diffuser, the effective band narrows significantly. Note that for an ideal diffuser, the band is limited by its physical dimensions, but for a different reason. The speed of sound in air has a final value of about 340 m/s at room temperature. At a certain frequency, the sound wavelength becomes comparable to the size of the diffuser and even smaller. In practice, this manifests itself as a narrowing of the directional pattern of the dynamic head with increasing frequency. That is, the higher the frequency, the closer to the axis of the head the listener must be in order to hear high frequencies. Thus, for a diffuser with a diameter of 10 inches (250 cm), the theoretical maximum frequency at which the acoustic radiation pattern is compressed into a narrow beam is 1335 Hz.

For the most commonly used size of 8 inches (200 mm) it will be already 2015 Hz, for a driver with a 5 inch (125 mm) cone - 3316 Hz, and for a typical tweeter with a diameter of 1 inch (25 mm) - 13680 Hz. At low and medium frequencies, designers try not to force the heads to work above these frequencies. For HF heads you have to resort to technical tricks. As a rule, a divider of one shape or another is installed in front of the diffuser, depending on the plane in which the radiation pattern needs to be expanded. In our example of an HF head design, a six-beam splitter provides optimal dispersion in both vertical and horizontal planes. In midrange heads, splitters in the form of cones with a complex generatrix are also used to expand the diagram.

A very important parameter of speakers is the linearity of its amplitude characteristics. This is the dependence of sound pressure on the amplitude of vibration of the diffuser. In a certain range of average values ​​everything works fine. However, at small values ​​of the input signal, the interaction force between the field of the coil and the permanent magnet is not enough to overcome the elastic forces of the suspension. This manifests itself audibly as a deterioration in the reproduction of low frequencies at low signal levels. At large amplitudes, the coil goes beyond the limits of the magnet field in the gap, which sharply increases the level nonlinear distortion. The amplitude of movement of the diffuser, within which the amplitude response of the head remains linear, is very small. For LF heads it rarely exceeds 6 mm, and for HF heads it is 0.3 mm. Due to such a small stroke, to improve heat transfer in the HF heads, the gap of the magnetic system is filled with magnetic fluid, which is a mixture of silicone grease and the finest powder of ferromagnetic material. However, their use limits the service life of the head due to a significant increase in lubricant viscosity over time.

Speaker selection remains the most important among other system components for the final sound you want in your listening room. On top of that, speaker systems have a very wide price range: from less than $100 to more than $70,000 per pair. The question arises what is inside if the price is so high. The answer is as simple as with expensive amplifiers. More expensive loudspeaker systems are produced in small quantities, they feature custom-made drivers (and, in addition, carefully selected parameters) and high-quality cabinets, most often handmade. In general, you see what you pay for, but the tonal characteristics of speaker systems are individual: the differences from sample to sample are perhaps greater than those of all other components of the sound reproduction system. You need to listen and listen to different systems to finally find the one whose sound is most pleasing to your ear. One speaker produces a bright sound in the highs, another produces a harsh sound in the mids, and the third produces very deep bass. Although, of course, there are systems with a more neutral (tonally correct) sound, there is no loudspeaker that correctly reproduces the entire sound range (the one that the human ear hears). They all color the sound to varying degrees, which depends on their price. Sometimes tonal coloring is specifically added to suit the taste of the speaker's creator. Finding speakers that suit your taste takes effort and time.

Alexey Grudinin (Stereo&Video)