The Development
of the DDD Driver
Truly Revolutionary
Award-winning Speakers
The German Physiks DDD driver is unique. It is an omnidirectional loudspeaker driver with an exceptionally wide frequency range and a uniform phase response. This text describes how the DDD driver works and compares it with the other most commonly used types of loudspeaker driver.
At first glance the DDD driver looks like the pistonic drivers that are used in the majority of the loudspeakers on the market. It has a voice coil/magnet assembly that serves as the actuator and it has a cone, though this is longer and narrower than usual. The shape is where the similarity with a pistonic driver ends.
The pistonic driver
With a pistonic driver, when the voice coil moves, the entire cone moves together with it – or that is what we want it to do. This is why the cone and voice coil structure is made as rigid as possible. Pistonic drivers are generally placed facing towards the listener, as excepting low frequencies, the sound waves generated are projected foreword, moving in the same direction as the cone - figure 1.
The DDD driver
Despite its apparently simple appearance, the way the DDD driver operates is considerably more complex. It has 3 modes of operation and in essence works as a mechanical 3-way system. These modes are:
Pistonic: This is employed at low frequencies up to about 200Hz. This is the same mode of operation as is employed in the pistonic drivers used in conventional loudspeakers, where the voice coil and the driver cone move back and forth together in unison.
Bending wave: The DDD driver cone is made from 0.15 mm thick carbon fibre sheet and unlike the cone in a pistonic driver, it is very flexible. As the frequency increases, the force exerted by the voice coil eventually becomes sufficient to cause the cone to flex and this in turn causes a wave to travel down the cone toward the open end. This is called a bending wave. Bending waves travelling in thin media such as the DDD driver’s carbon fibre cone, show a property called dispersion. This means that the velocity of the wave varies with frequency. In this case the velocity is proportional to the square root of the frequency. At about 200Hz the velocity of the wave in the cone wall reaches that of sound in air and at this point a sound wave starts to detach from the cone surface. Initially the angle of the sound wave to the cone, is very small, but this increases as the frequency increases, eventually reaching about 85 degrees.
The transition from radiation by pistonic motion to radiation by bending waves is progressive. As the frequency increases from 200Hz, the amount of energy radiated by pistonic motion gradually decreases, whilst the amount of energy radiated by bending waves gradually increases, until at about 800Hz, all the energy is radiated by bending waves in the cone. This change in operating mode is achieved smoothly, without any changes in phase.
Modal radiation: As the frequency further increases, the velocity of the bending wave in the cone increases and so its wavelength decreases. Eventually the wavelength of the bending wave equals the length of the cone and a standing wave is set up on the cone. This is known as the dipole frequency and for the carbon fibre cone, this occurs at about 4,000Hz. At this point the cone goes into break-up, however, rather than causing distortion as it would in a pistonic driver, this is exploited to allow the frequency response to be extended.
More about modal radiation
To give a very simplified explanation, when the DDD driver is radiating modally, patterns similar to the concentric rings of ripples you see when you drop a stone into water are established on the cone surface. Each one of these acts like an individual sound radiator. The small amount of cone material involved in each one means that the effective moving mass of each one is very small. As the frequency rises, the number of these individual radiators increases and thus their size and so their moving mass gets smaller.
By optimising the key properties of the cone material, i.e., thickness, elasticity and specific weight, together with the cone’s bending stiffness, the outputs from the three modes may be very closely balanced.
While the shape of the DDD driver’s cone gives it rigidity at rest, it is relatively easy to excite waves in the cone material. The clever part is controlling these waves. The resulting sound waves are radiated sideways from the driver through 360 degrees, as shown in figure 2. For this reason, the DDD driver is always mounted vertically.
Low moving mass increases frequency range
The bending wave and modal radiation modes cover the majority of the DDD driver’s frequency range and are what distinguish it from conventional drivers. In these two modes, when the voice coil moves, the whole cone does not move together with it, as it would with a pistonic driver. Instead, as mentioned above, the motion of the voice coil causes the cone to flex and a wave to travel from the top, down towards the open end. The voice coil therefore only has to move a mass that is a fraction of the total mass of the cone. This reduced moving mass allows the DDD driver’s frequency range to be extended up to about 30kHz, though in practice we start to roll it off at 24kHz.
The open end of the cone is terminated with a rubber suspension which is semi-rigidly attached to the chassis. This restricts the cone movement at low frequencies, so we normally start to roll off the low frequency response around at 200Hz, although on the German Physiks Unicorn Mk II, a single DDD driver coupled to a horn system covered the range from 40Hz to 24kHz. This is just one example of the exceptional performance offered by the DDD driver.
The exceptionally wide frequency range of the DDD driver allows the crossover point in the midrange that conventional loudspeakers must have to be eliminated and with it any problems this may cause are also eliminated. This is important because human hearing is most sensitive in the midrange, so even small problems here will have a detrimental effect on the sound quality.
Optimising the DDD Driver
The magnet
The DDD driver utilises a powerful ferrite magnet of our own design together with an under hung voice coil. The magnetic strength in the gap is approximately 1.2 Tesla. This gives the DDD driver a sensitivity that is on a par with that of conventional pistonic drivers of similar dimensions. The very high magnetic strength in the gap also provides a useful increase in efficiency in the upper two operating regions.
The linearity of the DDD driver’s magnetic circuit – a parameter that is crucial to the performance of any dynamic driver – is also extremely high. This ensures that the force that the voice coil produces to drive the cone very accurately follows the input audio signal.
To limit the moving mass, the weight of the voice coil must be minimised, consequently the upper frequency limit and maximum power handling become issues. In early prototypes, the upper frequency limitations posed quite a challenge.
Today we achieve a power handling capacity comparable with that of conventional woofers by using a voice coil constructed with flat wire that is wound on its edge, allowing an extremely densely packed winding and also by completely enclosing the coil within the magnetic structure, which then serves as a more efficient heat sink for the voice coil.
Controlling the Cone – Part 1
Minimising ringing
In any driver of this type, you want the motion of the cone to precisely follow the electrical signal that drives it. You do not want it to ring. This is where the cone continues to move after the impulse that excited it has finished. Essentially perfect control of the cone and an absence of ringing are easily achieved when the wavelength of the frequency propagated down the cone is greater than the length of the cone itself. It is difficult when the wavelength is shorter, since the full wave is reflected from the boundary of the cone – this is where the end of the cone opposite the voice coil attached to the driver chassis. The reflection will in turn produce re-reflections, as the travelling wave slowly loses energy over the course of several wave cycles. Imagine small ripples in a pool and how they are reflected back from its edge in a recurring pattern. The behaviour of a rippling cone is precisely the same, as the motion tends to persist for a considerable time, which ultimately has the effect of obscuring the information being reproduced.
Our approach to this problem is two pronged
Firstly, Peter Dicks, the designer of the original DDD driver, found that bending waves disperse, i.e., the DDD driver cone exhibits an increasing velocity of propagation of waves (more correctly termed phase velocity) with increasing frequency. He was able calculate the velocity of propagation in the cone as a function of frequency and therefore determine at what frequency the velocity of a wave in the cone reached that of sound in air. This is called the Coincidence Frequency. At this frequency the waves start to detach from the cone’s surface at an angle given by the ratio of cone-speed to air-speed. A few Hertz above the Coincidence Frequency the detachment angle is close to zero, but with higher frequencies the elevation approximates to 90°. The detachment of waves from the cone describes the functional principle of the bending wave radiator.
Secondly, he devised an effective means of balancing damping of the cone and the characteristics of the cone termination in order to minimise the residual ringing in the cone.
The calculations required to determine this were extremely involved, which is why to the best of our knowledge, no one prior to Peter Dicks had been able to devise and solve the necessary equations.
Controlling the Cone Part – 2
In addition to dispersion, the velocity of the bending wave in the cone is also affected its varying stiffness, which decreases as you move from the top, to the open end. Consequently, the velocity of a wave in the cone is proportional to frequency and inversely proportional to the distance from the cone’s peak.
The carbon fibre material we now use for the cone material and the titanium foil we used in the past, both have a high velocity of propagation in their bulk form. The velocity of propagation is especially high when they are formed into a very steep, thin-walled cone. Due to dispersion, the velocity of propagation of the wave in the cone is highest at the top, consequently, the upper part of the cone reaches the Coincidence Frequency much earlier than the lower part. Additionally, in the lower part, the wave lengths are shorter and all the waves are denser and therefore more efficiently radiated into the surrounding air. This higher efficiency of radiation means that there is less energy left in the cone to be reflected at the termination (i.e., the surround) and by maximising the amount of energy being radiated, a simpler form of termination for the cone may be used. This is important as the reflections would otherwise cause ringing in the cone. The theoretical way to eliminate this would be to ensure that all the remaining energy in the wave was absorbed by the termination, but this can only be done by having a termination that is the complex conjugate of the cone’s impedance and this cannot be physically realised.
With its current dimensions, the carbon fibre DDD driver reaches the Dipole Frequency below 4kHz. At this point the first standing wave starts to build up and above this frequency, the wavelengths become progressively shorter than the cone. Once beyond the Dipole Frequency, dispersion and radiation work as damping mechanisms to control the motion of the cone, however, additional damping may be employed to smooth the pulse and frequency response even more. These control ringing very effectively and are important factors in the carbon fibre DDD driver’s exceptionally extended high frequency response.
Minimising Doppler distortion
Another advantage of the DDD driver is that because it produces bending waves in a wide frequency range above the bass region, movement of the cone caused by bass signals produce almost no Doppler distortion, the bane of conventional loudspeakers.
The Free Lunch – with two courses
Omnidirectional radiation pattern
The DDD driver mounted vertically possesses an added advantage that is arguably just as great as the outstanding linearity of the design itself. That is, it is a nearly ideal point source with an omnidirectional radiation pattern.
The DDD driver propagates sound in a uniform hemispherical pattern. The frequency and phase responses are uniform from all listening angles, which is never the case with multi-way cone and dome loudspeakers, nor with dipole electrostatics or ribbons.
An omnidirectional radiator has several important audible advantages:
1. The window in which good stereo imaging can be perceived is considerably widened.
Unlike with a conventional loudspeaker, where you have to sit in a small area of the room to enjoy a good stereo image, with an omnidirectional loudspeaker, you can enjoy a good stereo image with the correct tonal balance in a very wide area within the room: just as you would in a concert hall.
2. The loudspeaker’s behaviour tends to be much more predictable from room to room, because the reflected sound is timbrally matched to that of the direct sound. Having the reflected and direct sound timbrally matched is also important for good stereo imaging.
3. The sound of an omnidirectional loudspeaker has decay characteristics more closely resembling large room reverberation than is the case with the narrowly focused output of typical monopole direct radiators. The sound has a naturalness about it that powerfully suggests the sense of space experienced at a live musical performance.
4. Because of the DDD driver’s physically compact size and exceptionally wide frequency range, it avoids the smearing of the stereo image that can occur when different parts of a musical instrument’s frequency range are reproduced by different types of driver, which are in different physical locations on the loudspeaker’s cabinet. This is because the distances between these drivers and the listener’s ears are not the same, which results in their contributions arriving at the listener’s ears at slightly different times.
Linear phase response
Of equal importance is the DDD driver’s uniform phase response. This is more often referred to as a linear phase response. A device that is linear phase, preserves the phase relationships of all the frequencies that pass through it. Music is a complex signal that can be broken down into a wide range of sine wave signals with specific frequencies and specific amplitudes, and very importantly, specific phase relationships between these frequencies. This mix is continuously varying. If an audio component does not accurately preserve these phase relationships, then its output will not be a faithful copy of its input.
Because the DDD driver is linear phase over its very wide frequency range, its tonal accuracy and transient response are both exceptionally good. Having a good transient response is especially important as music is a series of transients.
Why is the DDD Driver Truly Revolutionary?
To answer this question, we first need to look at the principal types of loudspeaker driver currently available. All loudspeaker drivers may be categorised by two groups of fundamental characteristics.
Group 1: This describes the means by which electrical energy representing the audio waveform is converted into mechanical energy and this in turn breaks down into three basic types.
Electromagnetic or electrodynamic
A simple reciprocating motor drives the cone, diaphragm or sound producing element. Well over 95% of loudspeaker drivers sold today fall into this class, which includes the DDD driver.
Electrostatic
Air motion is produced by varying the electrical charge on a membrane located between two electrodes.
Piezoelectric
These employs materials which flex in one dimension when an electric potential is applied across them.
While other methods such as magnetostriction and corona discharge modulation have on occasion been employed, none has achieved commercial significance, so the three categories above cover almost every loudspeaker driver currently available.
Both the electromagnetic and the electrostatic types can offer high levels of linearity. The piezoelectric type has generally been confined to applications where high sound quality is not a requirement.
Electrostatic loudspeakers are highly regarded by many audiophiles, and deservedly so on the basis of their sound quality at moderate output levels, but they suffer from a number of serious limitations:
Restricted maximum output level, especially at low frequencies.
Medium to low electrical efficiency.
High directivity, particularly at high frequencies.
Poor reliability especially in humid environments.
They can be difficult loads for some amplifiers.
We have done extensive research on electrostatic loudspeakers, but have concluded that they are inherently impractical and unlikely to progress much beyond their current state of development, consequently, the electromagnetic type is in our view the transducer of choice.
Group 2: This fundamental group of characteristics deals with how the mechanical energy is converted into acoustic energy and breaks down into the following types.
Mass loaded pistonic drivers.
Tympanic or membrane drivers.
Flat panel loudspeakers such as the NXT® and BMR® devices.
The Heil® air motion transformer.
Transmission line drivers, chiefly represented by the Jordan Module® and its developments; the Manger® driver; the Walsh® driver and the German Physiks DDD driver.
Mass Loaded Pistonic Drivers
This includes nearly all conventional cone and dome drivers. The name itself refers to the dominant effect of mass on the acoustic output of the driver, and to the theoretical model of pistonic motion to which such drivers conform to a greater or lesser degree. According to this model, the driver actuator and cone (usually a cone or dome), should move back and forth as a single unit within a single dimension, just like the piston in a reciprocating internal combustion engine. Ideally, the cone of the driver should remain entirely rigid and should exhibit no internal vibration whatsoever, though in practice this condition is never met, except at the lowest frequencies.
In such designs, mass reactance will be the main component of the complex acoustic impedance of the cone throughout most of the useful frequency range of the driver, hence the term, mass loaded. The mass in turn is loaded by the compliance or springiness of the driver’s suspension, and the two together, mass and compliance, form a resonant system like a weight suspended from a spring. Such a system tends to oscillate around a single frequency when excited, and, predictably, a large part of conventional loudspeaker design is taken up with damping such oscillations.
Unfortunately, a resonant system, even a damped resonant system, is poorly suited to sound reproduction.
Audible sound spans 10 octaves, while a strongly resonant system is mechanically efficient over only a small range of frequencies close to its resonant frequency. Such systems are necessarily bandwidth limited due to this frequency dependent efficiency, while transient response is inevitably degraded due to the inertia of the mass and because of the energy stored in the loudspeaker’s suspension and which is returned subsequently.
Such a driver, unless it is loaded into a horn or acoustic lens, will also suffer from a non-constant directivity pattern, with dispersion normally narrowing with increasing frequency. This is often called beaming. This characteristic, perhaps more than any other, will cause a loudspeaker to sound musically unnatural since acoustic musical instruments almost never radiate sound in this manner.
The normal method for dealing with the limitations of mass loaded pistons is to use two or three per speaker system and to drive them over the narrow frequency ranges around their respective resonant frequencies. Many fine speaker systems have been designed this way, but we believe this approach is fundamentally flawed, and nearly always results in location dependant transient response, ragged directivity patterns and an overall sense of individual drivers that are imperfectly integrated.
Tympanic or Membrane Drivers
This includes film and leaf transducers, electrostatic and ribbon type dynamic loudspeakers. All of these use a diaphragm that has a low mass and a large surface area and which is stretched over a frame: in other words, the diaphragm and suspension are one and the same. In such designs the resistance of air dominates the acoustical impedance of the diaphragm, except at the lowest frequencies. Consequently, transient response and frequency range can be excellent. These designs normally lack high power handling capabilities and are incapable of large excursions, so they are best confined to the higher frequency ranges and are impractical in bass applications. Moreover, they typically dictate a line source configuration, which makes them difficult to use in domestic listening environments. They also tend to be difficult to integrate with mass loaded pistons due to their very different acoustical characteristics. As a result, hybrid systems incorporating them very rarely provide satisfying results.
Flat Panel Loudspeakers
The NXT® driver can be described as a special membrane driver where its technology is based on the modal excitement of a stiff panel diaphragm, where a number of areas are stimulated, but using a single exciter.
The number, location and size of these areas vary with frequency. While optimising the location of the exciter point for a given panel may flatten the frequency response of the loudspeaker, it always lacks the phase coherence and wide dispersion of energy which are mandatory for high fidelity loudspeakers.
Balanced Modal Radiator
This technology, a spin off from NXT®, is based on pistonic movement. It uses a single exciter, but in contrast to the NXT® panel, only one vibrating area is employed and the size of this area varies inversely with the frequency of the stimulus.
The Heil Air Motion Transformer
This is an extremely ingenious design with high output, high efficiency, wide bandwidth, good impulse response and low distortion. Its sole real drawback is its low frequency limit - not much under 1kHz - and the consequent necessity of mating it with a conventional woofer. Sadly, most attempts at doing so have resulted in audible mismatches.
The Transmission Line Driver
This is the class that the DDD driver falls into, which when optimised offers what we believe to be currently the best overall performance in sound reproduction. Whilst similar in construction to a mass loaded pistonic driver because of its cone and use of a conventional voice coil and magnet type actuator, it differs in two major respects. The cone is securely anchored at its mouth and is flexed by the motions of the voice coil, and only pushed back and forth up to its Coincidence Frequency. From there on sound propagation is at angles up to 90 degrees to the wall of the cone, rather than parallel to the path of the voice coil, as is the case in the lower pistonic frequency range.
The cone itself ideally has an extremely high stiffness to mass ratio, but because it is very thin and its moving mass is extraordinarily low, so also is the bending resistance. As a result, the cone can be excited into bending modes quite easily, particularly when the velocity of the waves on the cone is higher than the velocity of sound in the surrounding air and consequently the wave energy will detach from the cone surface.
Without going into the physics of traverse wave propagation across a plate structure - which essentially is what the cone is in this design - we can say that when the cone is bent by an actuator, the actuator itself - in this case the voice coil - sees only a very small increment of mass from the cone. This gives an exceptionally low moving mass.
Rather than being mass loaded, it is loaded instead by the differential stiffness per specific weight of the cone-material and secondarily by the radiation resistance of the air load on the cone.
In simple terms, the voice coil is exciting shock waves across the surface of the cone which in turn excites motion in the air. As distinct from a conventional cone, there is almost no mechanical inertia to overcome, thus there is a very direct translation of the electron motion of the audio signal into the motion of air molecules in the listening space.
In a real sense, the acoustic behaviour of the system is much closer to that of an electrostatic membrane speaker than to a mass loaded cone, to which the transmission line driver bears a misleading external resemblance. The moving mass of the German Physiks DDD driver is under three grams, less than that of most tweeters, and yet its ability to displace air is roughly equivalent to that of a 6.5-inch woofer. So, while it shares the cone shaped diaphragm of the latter, its behaviour could not be more different.
Conclusion
To come back to the original question, why is the DDD driver so revolutionary?
At a stroke most of the limitations of conventional drivers have been eliminated. The combination of high displacement, low mass, and high acceleration allows the DDD driver to operate linearly over a very wide frequency range, nearly the whole of the audible spectrum and to achieve excellent impulse response, low distortion and a flat phase response into the bargain.
The German Physiks DDD driver is able to offer an improvement in sound reproduction that we feel quite justified in describing as revolutionary.
We acknowledge the following trademarks
Hawaphon® is the trademark of Korff AG, Switzerland. www.korff.ch | Jordan Module® is the trademark of E.J. Jordan Designs, UK. www.ejjordan.co.uk | Manger® is the trademark of Manger Products, Germany. www.manger-msw.com | NXT® and BMR® are the trademarks of FLAT Audio Technologies LLC, U.S.A. www.tectonicaudiolabs.com | TDL® is the trademark of Audio Partnership PLC., UK. www.audiopartnership.com | Walsh® is the trademark of Ohm Acoustics Corporation, USA. www.ohmspeakers.com.