The times we live in are not the times of a speaker revolution. The foundations of most of the solutions currently in use were invented many years ago and today are being improved, at the most. In the 1960s and 70s, so fruitful both for the music and music playing technology, there were tens of fresh ideas, including on original speaker sets – unique not for their visual but structural design. Today, the proportions are reversed – the main emphasis is placed on design as this is what the customers, supposedly increasingly often bullied by women they live with, whatever it should mean, expect. Also the quality of sound, however, with small steps forward that have been made for so many years has undoubtedly reached a higher level and became available for an affordable price. Even so, for professionals it is obvious that presenting an increasingly long list of “proprietary solutions” as inimitable, unique for a specific company and guaranteeing quality improvement impossible in any other equipment, is dictated solely by marketing objectives – and possible owing to the non-professional customers’ ignorance of the matter. Even such obvious characteristics of a good loudspeaker as properly aligned crossover network and a robust cabinet are sometimes being pointed out to customers as warranting greatest admiration.
ESA does not contend with specialists in devising marketing baits. For the fifteen years of our company’s existence we have been dealing nearly exclusively with designing speaker sets. We have been using conventional, proven and effective formulas. They require, however, not only using the best transducers but also their correct matching, mating organization, coordination with all structural features, with proportional consideration of primary and secondary issues. Nor should we accept the familiar schemes as absolutely binding. Any innovation, being a component of technological progress, is a successful attempt of breaking prior conventions. With time these innovations become standard solutions which, however, only await further changes…
The cabinet – its type, size, shape, all inner and outer measurements, enforcements, alignment method, attenuation; the electrical system – type of filters, grade of components, assembly method, cabling; finally, the transducers themselves… this is a very complex system of communicating vessels, closely related phenomena, a conglomerate of mechanical, magnetic, electrical, acoustic components. The secret of the ability to construct speaker sets consists in becoming acquainted with parameters of individual system components and relationships between them, as well as in reading their actual meaning. Every product, even declared to be “uncompromising”, is made within a specific budget and reasonable limits. In practice, then, compromise is necessary and the whole trick is to render it accurately. Day-dreaming leads to uncompromising solutions, existing only in the imagination and promises of the catalogues. Understanding the facts and costs leads to optimum solutions, feasible in the real world. The latest ESA constructions feature not only excellent Danish transducers and crossover network components, more than averagely robust and perfectly finished cabinets, but first and foremost they make maximum use of the components for final acoustic and visual effect. Following the track of relationships between individual features and components of a speaker system, we have obtained a really new, proprietary solution we have abbreviated to PPW – move, tilt, align (from Polish: Przesuń, Pochyl, Wyrównaj). We move the speakers, tilt the front wall and align the characteristics.
Linearity of processing characteristics is not the aim as such and is not the only parameter that the design engineer must bear in mind. If, however, natural sound, that is neutral processing, is desired then – undisputedly – good linearity of characteristics may be nothing but helpful. The characteristics usually presented – measured on the selected principal axis or possibly on a few other directions – show only a fragment of the problem of the general tonal equilibrium. Speaker sets radiate acoustic waves not only directly towards the listener but in all directions, however this radiation is far from regular – the acoustic pressure changes both with the distance of the listener from the principal axis of radiation as well as with the frequency. What reaches the listener is not only the wave directly from the speakers (though it has the greatest share) but also waves directed by the speaker in other directions and then reflected from various surfaces in the listening room. These surfaces are in various distances from the speaker and have various absorption characteristics, which means they attenuate various frequencies in various ways. The arrangement deciding about the tonal characteristics of the sound reaching the listener becomes even more complex, and simultaneously unique – each room has a different acoustics; moreover, in various sites within one room the same acoustics will yield different effects. Even if we were able to deliver each processing characteristics by free determination of the acoustic pressure value on any axis and at any frequency, it is impossible to clearly define theoretically what these characteristics should be like…
Hence the dramatic variety of concepts – speakers and speaker sets radiating in a very specific direction (horn loudspeaker) or in all directions (roughly, of course), or bidirectionally (bipolars and dipolars); narrow front walls allow greater dispersion (however at the cost of effectiveness measured on the principal axis, as a large proportion of the energy “runs away” to the sides and to the back), wider act just the opposite… In spite of all these conceptual divergences, or maybe due to them, most design engineers does not question the fact that the only relatively universal test of a speaker set’s tonal equilibrium is to measure the processing characteristics on the principal axis. This is the image of sound running directly from the speaker to the listener. If we are unable to determine the invariable principles of interaction between the speaker set and the room, we must completely disregard those phenomena and completely eliminate their effect, i.e. any reflection, to standardize the measurement method. During measurement these are eliminated by a dead room or an impulse system, which closes the measurement time before reflected waves make their way to the microphone. Obviously, these conditions are only a convention (and not a perfect one – sound in a dead room, though distinguished by a unique precision of localizing apparent sound sources, does not seem natural at all because a complete lack of reflections is not characteristic of natural acoustic environment in which music is created), but they are the only reasonable conditions for comparisons between different speaker sets, allowing to establish whether at least on the principal axis the loudspeaker is operating correctly.
In each room there is, however, one common factor strongly affecting the sound as it affects the pressure course in the listening site. This is an effect which can be predicted and taken into account in designing a speaker set. This is never done because measurements disregarding any reflections as presupposedly random (taking into account completely different rooms) also reject the effect of such reflections which occur in an easily predictable way. The effect of waves reflected by the floor! Freestanding loudspeakers, also called floorstanding loudspeakers, stand on the floor as their name indicates. Other surfaces – the ceiling, walls, windows, furniture – can be located at a greater or smaller distance but the floor is always exactly where the loudspeaker set is standing. And where the listener is sitting. In the case of stand mounted speaker sets, this issue is only slightly uncertain – the stands are usually 60 cm high and thus set the speaker’s distance from the floor. Unfortunately, our concept of correcting the errors caused by reflections from the floor is only practicable in floorstanding devices. Why do we call them “errors”, anyway? Reflections are after all supposed to have a positive effect on the naturalness of sound… That is correct, but does not apply to reflections from surfaces located so close to the source. Waves reflected from the floor will be delayed in the listening site compared to the direct waves to a too little degree to be able to make the sound “spatialized”, but enough to be able to distort the location of apparent sound sources, and reaching their target in a phase offset against the direct wave phase (they have a longer way to cover) they will interfere with it in the listening site, causing amplification of the acoustic pressure at some frequencies and damping at other.
We cannot find what frequencies these will be without establishing the distance between the speaker and the listening site. The speaker and the listening site define the location of the point on the floor in which reflection reaching the listener will occur (reflections will obviously occur on the entire floor but most of the reflected waves will by-pass the listening site), and then these three points make up a triangle in which the direct and reflected waves compete. When the listening site is moved, the triangle changes its shape, the ratio of its side lengths, and thus also phase relations. Even so, at a specific height – let us assume 80 cm – of the speaker (midwoofer, as this frequency range will be mostly reflected from the floor and distorted) and the height of the listener’s ears (let us assume 80 cm as well) it can be found that moving of the listening site from 2 to 5 meter distance changes the difference in the distance to cover for the reflected and direct waves from 56 cm to 26 cm approximately. This results in damping of the wave, half of which is of such length exactly. With approx. 2-meter distance between the speaker and the listener, damping occurs at approx. 300Hz, and with 5-meter distance at approx. 650Hz. There is no hope that any speaker will radiate waves of such length directionally, i.e. only towards the listening site (it is anyway easy to check how small is the angle to the principal axis of the direction leading to reflection from the floor, running further to the listening site – on the distance of 5 meters between the speaker and the listener it will be as small as 30°). Neither are there any chances for attenuating such waves even with a very thick carpet… We have made attempts, as this is one of procedures necessary for impulse system measurements with time frame extended at least as much as to be able to measure the 200-300Hz band; to markedly attenuate reflections from the floor, the floor needs to be covered with an attenuating structure several dozen centimeters thick.
fala bezpośrednia – direct wave
fala odbita – reflected wave
How can it be helped? In fact, with one midwoofer, i.e. in two-way structures – floorstanding or stand mounted – it cannot be helped. Purely theoretically, if we brought the midwoofer closer to the floor, the difference in distances for the direct and the reflected wave would be very small and would shift the phase problems to higher frequencies (which would also be easier to attenuate). Such a solution is not considered, though, as the midwoofer must be located directly by the tweeter, and the medium and high frequencies should be radiated from the height of approx. 1 meter (equivalent to the height of the listener’s ears – to obtain the impression of a natural location of the sound sources). Only the bass – due to its radiation all around and a very large proportion of reflected waves making source location impossible – may be radiated by any part of the speaker set (we make use of this phenomenon in the subwoofers). Certainly, many reservations and conditions required for correct operation of a woofer moved away from the midrange&tweeters can be presented here, but we only note the possibility of moving the woofer away from the other speakers. This way in three-way systems a very elegant solution is theoretically possible, a solution which would largely reduce the above-described problem of waves reflected from the floor (those reaching the listener).
If we take into account the conditions of the previous example – with listening point at the height of 80 cm, in the distance from the speaker set varying from 2 to 5 meters, but with the speaker at the height of 20 cm, then we can calculate that the waves running directly and those reflected from the floor will be in the opposite phase in the frequency range from 1kHz to 3kHz approximately. This is a range clearly above the frequencies usually processed by woofers (and also radiated in a much more directional way). If, then, in a structure with a midrange speaker at the height of 80 cm, and thus potentially capable of causing problems in the range of 300-650Hz, a crossover frequency would be set between 650Hz and 1kHz we would prevent the distortions resulting from reflections from the floor. Such a high crossover frequency, however, especially with such a marked shift, is not good for other reasons so this plan is rather out of the question. It is always worth to take these phenomena into account and both move the woofer closer to the floor and not enforce too low crossover frequency. Even if it is not in the range ideal for the purpose of solving the reflection problem, even a close neighbourhood will bring benefits. After all a woofer affects the characteristics also slightly above the crossover frequency and may, at least partly, compensate for the adverse phenomena caused by the midrange speaker.
This is why in the three-way Credo 4 the woofer section is moved away from the midrange speaker, and the crossover frequency is set not as low as the large resistance of the latter would allow (the maximum amplitude of which is more resembling of a midwoofer and not a midrange speaker). Also, always the smaller the power supplied to the midrange speaker, the smaller the resulting compression (created because of the coil temperature increase). Woofers are much better prepared to operate not only with high amplitudes but also with greater heat doses (which do not decrease so fast as the amplitude with the frequency increase). Woofer coils are not only longer (allowing a greater amplitude) but also larger in diameter, so have a much larger area resulting in much smaller temperature increase as a result of a specific heat power dose. To culminate this issue we should add that the woofers used have coils with titanium carcasses (hence with a very large heat capacity) being excellent radiators for the coil winding itself.
The main area of our study on reflection effect reduction, however, was the two-and-half-way systems. This is obviously also because this type of systems is much more popular nowadays than the three-way ones, and not without a reason. Moreover, the two-and-half-way system has proved very susceptible to our concept and against all appearances not at all less appropriate than the three-way system. We have shown above that it is difficult to completely evade the adverse reflections caused by the midrange unit – they are located to high to set the crossover frequency at an even higher level. Let us assume that the adverse effect occurs at approx. 500Hz (the speaker at the height of 80 cm, the listener – 100 cm, distance – 4 meters). If the midrange tones processing unit is trapped anyhow, then it can be considered a midwoofer from the point of view of reflections from the floor. This will not result in further adverse phenomena below 500Hz. However the addition of a woofer allows reduction of the damping at 500Hz which would be evident in operation of a two-way system – unless this speaker is located directly below the midwoofer as it is in most of the two-and-half way systems! In such a typical configuration, the adverse reflections occur for similar frequencies and are cumulated on the entire set’s processing characteristics. If the woofer at all reaches this range in processing. One way or another, the problem is unresolved even though an additional speaker allows it. What must be done is “only” to lower its position, not necessarily to 20 cm as we do not want its potential problems to be shifted much above 1kHz. Enough if they occur clearly somewhere else on the frequency scale than those caused by the midwoofer. A woofer must also process (not necessarily with full effectiveness) the range up to 500Hz to compensate for the damping caused by the midwoofer. Many measuring and listening experiments with settings and filters, with various distances between speaker sets and the listener (within 2-5 meters) led to definition of a similar configuration for all two-and-half-way systems: a midwoofer is typically set at the height of 80 cm, with tweeter directly above it and the woofer at 30 cm.
The filtering method for such a system is by no means trivial. Against appearances, development of a good crossover network for all two-and-half-way systems does not consist only in adding the woofer to amplify and better spread the low frequencies. One must take into account that the low-pass filter of the woofer, aligned to a lower boundary frequency than the filter for a midwoofer, will not only result in different processing characteristics but also different phase characteristics. And too large differences between the phases of these two speakers can lead to wave damping and further non-uniformity of characteristics. What must be done, then, is to synchronize phase characteristics of woofer and midwoofer sections. The simplest solution is to use a lower class filter in the woofer (which causes smaller phase shift in the stopband and directly before it) than in the midwoofer. Such a solution is also at stake in our concept. If, however, with mild filtering of the woofer we spread its characteristics to as much as approx. 700Hz (6dB drop) to make it compensate for the distorted characteristics of the midwoofer, then with effective mating of both speakers up to this frequency and simultaneous moving the woofer away we must take into account another adverse phenomenon – and, obviously, find a smart solution to it yet again…
At the vertical front wall, the speaker location defined above makes it seem to the listener at the height of 1 meter being not far away from the speaker set (let us say 2 meters away) that the woofer is 12 cm farther away than the midwoofer. This is nearly ¼ of the 1300Hz frequency wavelength, which means a phase shift of approx. 90° and in itself is not yet disastrous (two equivalent vectors set at an angle of 90° give a resultant vector still stronger than each one of them). We must bear in mind, however, that at this frequency, even with mild filtering of the woofer, we obtain phase shift caused by this filter larger than that coming from the midwoofer filter. Both phase shifts – from the previously attenuating filter and mechanic distance – regrettably go in the same direction and thus the resultant acoustic phase shift at approximately 1300Hz may approach 180° and hence cause wave damping. Therefore, this shift needs to be reduced by tilting the front wall, which should equalize the distances from both speakers to the listening site to a sufficient degree (depending on the distance and height the listener is located at, this difference may amount to up to 3 cm, which already is an insignificant phase shift of a 700Hz wave).
We have asked ourselves whether moving the woofer participating in 700Hz processing, 50 cm away would not result in blurring the soundstage due to the spreading of apparent sources of midrange and high frequencies. But even if both speakers sounded with the same frequency range producing identical sound intensity, the apparent sound source of midrange frequencies would be shifted by 25 cm – between both speakers. In our system, meanwhile, at 700Hz the lower speaker is damped by 6dB, thus the apparent sound source for this frequency is located much closer to the midwoofer. Going down the frequency range, the sound level produced by the woofer gradually approaches the level produced by the midwoofer (at 500Hz the difference is 3dB); the apparent source may shift but below 300Hz what we have is a delocation effect for apparent sound sources due to their all-directional radiation and reflections. It is also worth checking that the typical relation of distance between the midwoofer and the tweeter (15 cm) to the wavelength of the crossover frequency between them (let us assume 3kHz) is by no means smaller than our relation of distance between the woofer and the midwoofer (50cm) to the 700kHz wave. This means that neither will the directional characteristics in vertical plane be impaired due to this configuration. Thus, first the problem with phase interferences will arise between the midwoofer and the tweeter. This happens not only in theory, but also in practice – i.e. in intense listening tests in which such systems, upon final alignment, sounded with balance, cohesion and plasticity. The recommended distance of the listener from the speakers is within 3-6 meters.
Tilting of the front wall in our devices is therefore somewhat related to operation of 1st order filters (woofer filtering), equalization of the distance from the speakers and compensation of phase shifts, but has not arisen from the concept of a speaker set to feature a “linear phase” but from the above-described idea of mating the woofers and midwoofers in order to equalize processing characteristics in the mid bass range. As for organization of mating of the midwoofer with the tweeter, we are not convinced about the 1st order filters which make it very difficult to obtain equalized processing characteristics, deteriorate directional characteristics, burden the tweeter and increase distortions, and rarely achieve their aim – a phase characteristics giving perfect impulse response. Certainly, when the front wall is already tilted and the location of the tweeter in relation to the midwoofer has been adjusted, this will affect the crossover network shape, although it does not need to lead to using 1st order filters but can be used otherwise. Owing to this partial compensation of the phase shift between the midwoofer and the tweeter we can increase filter slope, keeping at the same time compatible polarization of all sections favoured by us.
Our new designs not only have the front walls but also the back walls tilted. This is yet another example of the interdependence of various characteristics of a complex system typical for our designs. We have made use of the necessity to tilt the front wall (for coordination of the woofer and midwoofer phases) to construct a cabinet of a shape best also in quite other respect – in reducing standing waves in the cabinet. Many myths have been said on this subject. A slightly tilted upper wall or rounded sides are enough to boast that the problem has been solved. The standing waves, meanwhile, are still raging as they will not subject to design and marketing solutions only. They are subject to different rules than many design engineers would wish them to. We have ourselves become humble upon our experience with many earlier models in which in the cabinet bottom we had installed a partition at an angle of 45° thinking this was a nearly perfect way of removing standing waves generated between the top and bottom wall. In spite of this, the bass-reflex radiation manner (which, apart from the basic system resonance and tunnel parasitic resonance, to some extent also transmits the standing waves from the cabinet) signalized that the problem still existed… It is commonly accepted that the standing waves are a result of parallel cabinet walls. Therefore, to remove these parallelism should be to remove the standing waves. Unfortunately, this is not so simple. Tunnel resonances (pipe resonances) arise in systems in which one dimension is much longer than the others – e.g. in a speaker with a height much greater than the width and depth. Then it is not enough to set the bottom or top wall even at an angle of 45°. It will not eliminate the marked privilege of one, vertical dimension of the device. If, however, the high cabinet with tilted front and back walls should be divided inside with a partition set at an angle of 45°, the situation would be much different. In the lateral vertical section not only there will be no parallel walls, but also the distances between them will clearly change. The only potential generator of standing waves is still the system of parallel side walls with the smallest distance compared to the distances between the other walls, which reduces the resonance risk and also complies with the wavelengths which can be handled using appropriate attenuation of the wall surfaces only.
Our devices have for many years been equipped only with excellent Danish transducers. The PPW generation is based on Scan-Speak’s reference products – the Revelators in Credo models, and the latest Peerless HDS series in Intrada and Triton models. The new HDSs have nomex cones, and traditionally for the Danish products are based on cellulose, this time enriched with new admixtures setting optimum parameters of rigidity and internal attenuation. 18cm HDS Exclusive midwoofers (in the Triton model) have magnetic circuits equipped with magnetic field symmetrization systems similar to Scan-Speak’s Symmetric Drive. The tweeter in the Intrada 3 model is a ring-dome DX tweeter, and in the Triton it is the already famous ring radiator XT (both tweeters have been developed by Vifa, and are now manufactured by Peerless). Selection of speakers is dictated not only by the price and rating of the models but also their matching – e.g. an 18cm Exclusive in the Triton models requires a slightly lower crossover frequency and therefore an XT is a better match than a DX. DX, on the other hand, with its excellent dispersion in the top octave perfectly matches the saturated and yet open sound of the Intrada midwoofer. The Triton models are more focussed, precise and neutral. Finally, Credo models add to this the most sophisticated, rich but free from exaggeration sound in the entire band. What can be expected of a well applied 28mm silk “large” Revelator 9900 we could see from numerous hi-end designs, not only of our making. The Revelator tweeter used in Credo 3 finally has a perfect and worthy companion – an 18-cm Revelator midwoofer. Each design detail is carefully touched up. Moulded arms of the cast basket, free ventilation of the coil, non-resonating lower spring (bellows of non-identical folds), notched membrane (each notch is manually coated), Symmetric Drive system, obviously, which here primarily guarantees a uniform and symmetrical distribution of the magnetic field, and thus very low distortions, also at high control levels. In the three-way Credo 4, the achievement of the reference level of low tone definition was helped by using a pair of 22cm alucone Revelators. Midrange is reproduced by a papercone 15cm Revelator and high frequencies – by R29, top ring radiator tweeter from Scan-Speak.
In terms of basic electrical parameters, our new speakers are typical, by no means problematic burden for each “robust” amplifier. Although most manufacturers declares an 8-ohm nominal impedance with impedance characteristics showing drops to as little as below 3 ohms in the range of low frequencies, in our devices we never go down below 3.5 ohms. At the same time, in compliance with the traditional and reasonable method of determining nominal impedance, we declare our speakers to have a 4-ohm impedance. Likewise with the effectiveness – we specify a reliable value determined by way of measurement in an open space. As with any high class device, also our loudspeakers play better connected to a better amplifier and player. However, both because of the nature of the transducers alone as of the sound profiling of the whole speakers, they are not very demanding. It will not be necessary to find the one and only equipment configuration or to buy particularly expensive devices. Clearly, however, they like good and well recorded music – bad music will sound badly