3 WAY HIGH EFFICIENCY HORN SPEAKERS
When designed and implemented correctly, horns can attain a level of reproduction and dynamics unmatched by any other type of speaker
"Years ago as a young audio enthusiast, I was fascinated with horn speaker technology, and quickly recognised the innate potential in their design. After a lot of reading and research of the physics involved I began to develop an appreciation of the prior art, while at the same time noticing how some contemporary horn designs had fundamental flaws that undermine the advantages and introduce additional problems. I wanted to take the concepts from the early horn era and push them further, taking advantage of the technology of today in terms of computing, simulation, CAD-CAM and manufacturing advancements, and use it to advance horn performance whilst retaining classic horn virtues and sonic performance."
DESIGNER
The Symphonia is the result of an intensive five year research and development program in horn speaker engineering
SYMPHONIA AERIS WITH LIQUID COPPER FINISH
The unique shaped tweeter horn, is a Zero Diffraction horn, with a 360 degree continuous flare, eliminating all possible diffraction in vertical and horizontal axis, delivering perfect performance in frequency and time domain, with unmatched transient and distortion figures, by any other HF transducer technology used.
The horn body is made from solid wood.
The development of the Symphonia’s tweeter was spread over a three year period. I was always mesmerised by the sound of a true ribbon, especially Raal’s designs, but usually direct radiating ribbons lack the punch and dynamics of a high-frequency compression driver. On the other hand, no compression driver could ever deliver the detail, accuracy and bandwidth of a true ribbon transducer. So we began research on whether it was possible to combine the two approaches to get the best of both worlds.
A normal ribbon is not designed to operate at high acoustic pressures typically present at the throat of a horn. It soon became clear a special ribbon would need to be designed. After modification and experimentation on the Raal tweeter, we found the end result to more than meet our expectations. However, horn loading the ribbon still presented challenges. We were reluctant to use conventional solutions such as wave guides or directivity control apparatuses that would be inadequate for the task, so we endeavoured to find a new approach.
This lead us to design a horn flare that specifically loaded the ribbon as a true horn. Nevertheless, the line source emission characteristic of the ribbon made this extremely difficult. Diffraction is secondary sound emission, and takes place in geometric anomalies, transitions of geometry and changes in curvature. It’s very much audible and almost impossible to eliminate. In horns, diffraction is present at the horn termination for round horns. In rectangular horns the problem is more pronounced.
The solution was the implementation of our diffraction-less 360 degree horn technology. We designed the horn flare to expand to 360 degrees, while maintaining the flare curve’s second derivative constant, keeping the curvature slope constant. Precise measurements were made of each iteration in an attempt to match the simulation and theory to its real-world acoustic performance.
We found variations of just 0.2mm to have detrimental effects on the horn’s performance. Manufacturing them to specs from wood was also challenging. In the end, a total of fifteen different flares were designed, built, measured and eventually discarded before we settled on the final design that performed as we wanted.
The ribbon horn combination response is shaped to measure flat from 3KHz upwards. It delivers detail and resolution associated with only the very best ribbons, with the solid punch and dynamics of the very best high frequency compression drivers. Distortion characteristics at 114db are lower that those of the best dome tweeters at 95db. What’s more, the diffraction-less nature of the horn enclosure gives the tweeter unmatched time domain performance.
The driver used is a large format (4 inch diaphragm) titanium driver, with a massive magnetic system (11kg magnet weight).
The midrange acoustic level can be altered via a multi-tap transformer, for tuning the speaker according to each room’s acoustics.
One of the things we wanted to overcome was the problem of diffraction that almost every horn suffers from. Diffraction is one of the major sources of so-called “horn sound”, whereas a true horn properly designed and implemented should not and will not have an identifiable sound of its own.
We decided to employ a round tractrix horn, modified with a specially-designed 360-degree diffraction-less flare. By using a bullet-shaped enclosure, we were able to completely eliminate colouration due to diffraction, and avoid the smearing of transient information that usually occurs.
We wanted the horn to have a cut-off at least 0.5 octave lower than the crossover frequency, so we designed it to have a cut-off at 330Hz. The horn and bullet-shaped enclosure is CNC machined from solid plywood block, with a thickness of up to 15cm. The poplar plywood used was preferred to other types of plywood, as this had twice the strength-to-weight ratio of other types of plywood. Energy storage is thus halved.
The driver used is a large format titanium driver with a 4-inch diaphragm, coupled to a massive magnetic system weighing 11kg. The midrange acoustic level can be altered via a multi-tap transformer, allowing the Symphonia to be better integrated according to each room’s individual acoustics.
Instead, we chose to employ a tractrix flare profile, strictly avoiding any sharp corners, deviation from actual theoretical geometry and with no abrupt curvature changes. This is a much greater challenge from an engineering perspective. Fortunately, we have a five-axis CNC router with a 5m3 work envelope that gave us the capability to design and manufacture a true tractrix horn from twenty-five vertically stacked sheets of plywood that satisfied all the above criteria. Another major advantage conferred by the stacked plywood structure was that is has the requisite mechanical rigidity to withstand the high acoustic pressure present inside the mouth of the horn, especially at these frequencies.
We believe that bass delivered from horn loaded drivers are unmatched in performance, as long as the horn in properly designed. Folded horns suffer from three major factors:
• Usually to short in length for the bandwidth used. Shorter horns and small mouth-area bass-horns, have large frequency response ripples, and strong coloration.
• The horn is often not folded correctly, the horn flare is formed by conical sections of wooden sheets, with sharp angles and many parallel surfaces, trying to resemble something like a true horn flare.
• Horns and bass horns in particular, submit their internal walls to extreme pressures. Plain sheets of wood are just not appropriate for this task.
The Bass horn used is a 3.3m long horn with very large mouth area for the bandwidth used, providing excellent performance in bass extension and weight, with the lowest coloration than any other folded horn.
The horn flare, is a true tractrix flare, formed by stacked sheets of plywood, carved to formed a true 3.3m long tractrix horn, properly folded without any internal sharp angles and geometrical deviation from a the tractrix flare.
The stacked plywood construction, provides 10 times the stiffness (along the horn axis), compared to conical horns formed by sheets of wood, also constructed with much reduced weight/stiffness ratio, providing minimum energy storage.
External Crossovers
The external crossover, consists of three constant group delay filters. A constant group delay filter, is our own proprietary design, which alters the attenuation slopes of the driver filters, to shape the phase (and doing so the group delay) in such a way as to provide maximum group delay linearity.
Using this specific filter technology to the physically time aligned horns, the Symphonia speaker present time domain performance which is near perfect, and a frequency response with a maximum deviation of +-0.5db (250Hz-20KHz).
ARIES CERAT SYSTEM | IANUS GEMINEA MONOBLOCK AMPLIFIERS, INCITO S PREAMPLIFIER, SYMPHONIA SPEAKERS
Technical Specifications
3 WAY HORN SPEAKER SYSTEM | |
High frequency transducer: | RAAL Ribbon tweeter with modifications. Housed in a 360° flare. CNC machined solid wood housing. Lateral adjustment. |
Midrange Horn Unit: | 4 inch compression driver horn, oversized magnets and mounted on a custom throat. Horn enclosure CNC machined in solid plywood. |
Bass Unit: | Tractrix horn loaded enclosure made of dense plywood sheets. Custom internal 3.3 metre horn exiting to a large mouth. |
Crossover: | Custom built 3 way constant group delay filters with time domain technology and filtering. Adjustable gain for the 3 way for room tuning. |
Sensitivity: | 101dB |
Frequency response: | (-3db) 32Hz~100Khz +-0.5db at 250Hz-20KHz |
Weight: | 170 kilos per channel |
Overall dimensions: | 930mm W X 1400mm H X 880mm D |
Paint finishes Aeris version: | Black with Bronze Metalic for the horns. Wood veneer for the bass unit sides. |
Paint finishes Custom Aeris version: | Liquid Copper finish for the horns. Crossover finish to match. |
CREATION OF THE SYMPHONIA - OUR JOURNEY
What lead to the idea to build a horn speaker like the Symphonia?
The Symphonia began as a personal project, long before the establishment of Aries Cerat in 2010. Shortly after we started the company, we initiated two parallel development platforms, the Contendo and the Symphonia. This gave us two horn speaker projects, allowing us to exchange concepts and innovations between them. The horn-loaded ribbon tweeter, for example, with its innovative 360deg flare to eliminate diffraction was initially developed for the Contendo, but was utilized in the final design of the Symphonia. Similarly, the solid wood midrange horn with its special bullet shaped enclosure – again, to eliminate diffraction – was developed for the Symphonia and then used in the new Contendo.
So, you can say that the Symphonia was not an isolated project, but a project that grew along with the company and adopted many breakthroughs from other speaker projects that were also in progress. Our proprietary variable slope filters with constant delay characteristics, for example, were in development for many years, and although our smaller speaker got to use them first, the development platform was the Symphonia.
Part of the reason for the Symphonia’s extended development was that a lot of work and effort went into eliminating problems that accompany horn physics, which if not addressed thoroughly produce mediocre results. When overlooked, these problems manifest themselves as nasty sonic traits that many – wrongfully – attribute to horns in general. In my opinion, a poorly designed horn will always sound worse than a poorly designed conventional speaker. If the design of a horn speaker is compromised from the beginning, either from budget limitations, technical resource limitations, or simply lack of proper understanding of the physics involved, horns are better avoided and conventional dynamic driver speakers should be pursued instead.
However, I still regard horns as the ultimate electro-acoustic transducer, but only when the specific aspects of horn design are respected and implemented correctly. This is what we tried to do with the Symphonia, and we strongly believe we’ve succeeded.
What are the major differences between historical and current design in horn speakers?
I don’t think it’s right to put all horn designs in one basket. Many of the old designs are landmarks of audio reproduction and still relevant in showing the new guys how it’s done. Of course, some of the older designs were very much flawed and extremely colored by today’s standards, so one must be careful and evaluate each design on its own merits. I find the problems of many contemporary horn designs stem from two main sources:
Firstly, the lifestyle-oriented design of most horn speakers. Unfortunately, many design choices prioritize aesthetics over function. That’s not to say physical beauty is irrelevant, but in-and-of-itself can never make up for compromises that negatively impact reproduction.
Secondly, a basic lack of knowledge of horn physics. Past speaker engineers had different priorities, and although they lacked access to modern CAD-CAM systems, simulation programs etc, they arrived at designs that respected the basic and advanced aspects of horns and always managed to take a step forward.
While acknowledging everyone will approach the inherent problems of horn design with their own ideas, I find it difficult to understand some contemporary designs that clearly ignore the basic principles underlying all horns, despite the historical knowledge and modern technology available.
What does it take to achieve a full range, full-spectrum horn sound?
To answer this, we have to differentiate a true horn speaker design from speakers than use waveguides or other directivity control apparatuses, hybrid designs etc. Unfortunately, categorization of a speaker as a “horn” often blurs the definition creating confusion in the audio world. Horn loading is something very specific in terms of physics, implementation and end result.
A horn is an acoustic impedance transformer. Using an analogy in mechanical terms, imagine you have a 1hp servo motor coupled to a feather-like axial load. This is analogous to a direct radiating driver. Energy transfer of such a system is very poor, mechanical energy is wasted and efficiency is very low. On the opposite extreme, imagine you have a 0.1hp servo motor coupled to a 1m diameter steel disk. Again, we have extremely low energy transfer, but for opposite reasons. This is usually the case with electrostats and planars.
For our purposes of maximizing energy transfer in the conversion of electromagnetic energy to acoustic energy, ideally the motor “impedance” must be equal to the load “impedance”. Again, using a mechanical analogy, a horn is simply the acoustical equivalent of a gearbox, matching the impedance of the load to the output impedance of the motor.
However, like most things audio, there’s no free lunch. Horns are bandwidth limited – you sacrifice bandwidth for efficiency. If we’re talking about reproducing the full audio spectrum, this inevitably means separate, specially designed horns intended for specific bandwidths. We’ve found that in theory, as well as from real world data we’ve collated, horns perform better when handling a maximum of 1.5 – 2.5 octaves, even if the drivers loaded within the horns can handle much wider bandwidth.
To achieve a full spectrum horn sound, you have to respect the horn physics involved. For example, let’s consider just the first two octaves from 27.5Hz to 110Hz. To get usable extension with bass horns, you need a certain length and mouth area for a given cutoff. Horn cut-off frequency is always a function of size. The goal for a compact, delicately proportioned speaker suddenly goes out of the window. It’s very difficult to design properly within those constrains.
The reality is you can’t bend the laws of physics in an attempt to get around cut-off frequency, hence why a lot of hybrid and semi-active designs exist in the market. We understand the appeal, but we feel the inherent trade-offs involved ultimately inhibit the design and reduce performance. In order to achieve a full-spectrum design over the entire audio spectrum with limited bandwidth horns, the only solution in pursuing a line of least compromise is multiple dedicated drivers and horns, and as an inevitable consequence, an increase in cost, size, weight and complexity.
How does the Symphonia differ from the other contemporary horns, and what’s needed to show the real potential of a horn speaker?
For the Symphonia project, we decided to address many of the underlying issues associated with horn theory that in practice are often severely compromised in lesser designs. We attempted to find the very best solution to each problem, with as little compromise as possible, and combine them into a single entity.
We took no shortcuts in our efforts to realise each proof of concept, whether relating to the diffraction-less horn enclosure, specialised group delay filters, bass horn design or the tweeter design. We wanted the Symphonia to be a platform to show how a horn should sound when horn physics are properly understood and implemented without compromise, eventually materialising in a design that’s small enough to go through doors – although only when disassembled!
When properly designed and implemented, horns can be a true magnifying glass of the components that precede them. That means they’re not very forgiving of lesser amplifiers and sources. Ambience, detail retrieval, micro dynamics and timbre are all encoded within and below the level of an instrument’s fundamental notes, and because of a horn’s high sensitivity and resolution even at normal listening levels, the first few milliwatts of power matter.
Even for peak signals, a horn will often be drawing only a few watts of power, so the need for signal integrity from the amplifier at low power becomes exponentially more significant. This is where solid-state and most push-pull designs fail – the first few milliwatts. Of course, heavy and inefficient speakers mask the artifacts inherent in those first few watts, and are usually coupled with amplifiers in which their low power behavior is therefore not revealed to be severely compromised in ways a properly designed and implemented horn will.
How would you compare the Symphonia with other horn manufacturers.Such as Avantgarde Acoustics or Cessaro for example?
There are many ways to design a horn speaker. But that’s not the same as saying all horn designs understand and respect the laws of physics. There are some very specific concepts that need to be addressed and solutions implemented if one is to avoid a compromised result. In conceiving the Symphonia we looked into the myriad of problems a horn design entails, and addressed each one with the least compromise possible. Four of the major problems associated with horn design include:
Too wide a bandwidth for a specific horn/Improper flare for the bandwidth
Many designs usually get this very wrong. Any horn will have a specific bandwidth within which it can work optimally. Usually, designers “stretch” the horn’s pass-band in an attempt to avoid having to use more horns/drivers for a particular design, making for a design that’s more compact, aesthetically appealing, is easier to manufacture, less complex, etc. The problem associated with a horn operating in the wrong bandwidth is that it produces high-order distortions as well as nonlinear distortions.
Horns are very sensitive to their termination geometry – i.e., the geometry that follows after the flare ends. Discontinuities in the geometry and even changes in curvature can generate a secondary emission. This delayed emission returns inward to the horn throat and driver, as well as radiating outward towards the listener. This is called horn diffraction and is very audible.
Horns installed on baffles present high levels of diffraction, while rectangular horns present high-order distortions and many diffractive surfaces. Diffraction is the major factor in creating the “honking” sound associated with some horns.
We believe tractrix flares have the best measurable and subjective performance. We only use round tractrix horns, modified to our 360-degree flare to eliminate diffraction. Both the tweeter and midrange enclosures use this technology. Unfortunately, these bullet shaped 360-degree horn flares are extremely difficult to design and build, and would be impossible to manufacture without extensive use of our CNC workstations. However, in my opinion, it’s completely worth the time and energy, as it solves the ongoing problem that seems to exist in other designs.
Improper driver used
Not all transducers are suited for horn loading. Many times improper drivers misbehave in nonlinear and unpredictable ways when loaded with a horn, irrespective of the horn’s design. Appropriate drivers are usually expensive and hard to acquire. The most important aspect is in finding the best driver for the intended bandwidth. However, this is never straightforward, and many parameters must be looked into, not just the more obvious ones found in textbooks.
Inferior horn construction/material
A horn presents sound pressure levels many times greater than the pressure present in a conventional speaker. A plastic horn, however thick it may be, will ring and be a secondary source of resonance, colouring the sound. A thin plastic one, no matter how aesthetically pleasing, is a complete failure in this regard.
In designing the Symphonia midrange horn, we conducted extensive objective and subjective testing, alternating between a 2.5cm thick composite 330Hz tractrix horn and an identical horn built from stacked poplar plywood with 15cm thick walls. Both our objective analysis using burst decay and listening tests revealed that the plywood horn was vastly superior, even though the composite horn at 2.5cm is still much thicker than the market’s standard.
Of course you can’t make 15cm thick plywood from injection moulding, so they’re constructed in a labour intensive process using CNC machining to extremely high tolerances. Many of our ideas and concepts could only be materialized after the acquiring of highly specialized equipment, such as our 5-axis work station with the highest Z-axis envelope available, making it possible to carve anything from our material of preference. Of course, this ends up being much more costly and time-consuming to manufacture, but presents a solution with the best possible performance.
Improper time alignment/filtering
Flush placement of the front of the horns so that they’re physically aligned might result in an aesthetically pleasing speaker, but the result is an incoherent sound – one of the many sonic maladies horns are accused of. Improper filtering of the horns is also a source for coloration. Crossover design and physical alignment must address all issues that derive from both the physical position of the horns as well as the actual acoustic filtering inherent in a horn-loaded driver. Solving these issues must be done in a way that avoids treating them as isolated variables – they must be treated as a singular one. This is rarely addressed.
Time alignment is not just physically aligning the driver’s diaphragms, as is very often claimed for speakers that call themselves “time aligned”. The acoustic centre of the driver is not often located at the centre of the physical diaphragm itself, instead it maybe be offset forward or behind the centre of the diaphragm and is often frequency dependant. During development, we time-align the drivers using customised measuring rigs that excite the filter-driver-horn as a complete entity, and only after analysing the transient response of each filter-driver-horn system do we physically align the horns relative to one another to be truly time-aligned in the acoustic domain. Visually aligning the driver’s diaphragms is of no use from an acoustic perspective.
The external crossover designed for the Symphonia consists of three constant group delay filters. The constant group delay filter is our own proprietary design. The frequency response of the filters at the stopband does not follow standard profiles used in speaker filters, neither passive nor active. Instead, the filter’s attenuation slopes are varied and shaped in a way that the phase derivative – the group delay – is kept both constant and linear.
Using this specific filter technology in addition to the true time alignment of the drivers, the Symphonia is able to produce time domain performance that is near perfect, combined with a frequency response with a maximum deviation of +-0.5db from 250Hz-20KHz.
Why do horn manufacturers often exaggerate claimed sensitivity ratings?
Overly optimistic sensitivity rating is observed not only in horn speaker design, but in most speakers, regardless of speaker technology. Unfortunately marketing departments think they can “push” the rating up by a couple of dBs, possibly hoping to gain some sales for owners of smaller amplifiers. Some very-well known speakers actually measure 6db lower than their rated sensitivity. However, it’s quite easy to spot and point out designs that exaggerate claimed sensitivity, simply by analyzing the design, drivers etc. Rating sensitivity is not a race – we think it’s better to just state the actual rating for real world performance comparison.