Diamond loudspeaker cones for high

Diamond loudspeaker cones for high-end audio components
E. Wörner1*, C. Wild1, W. Müller-Sebert1, P. Koidl1 and A. Bankewitz2
Fraunhofer-Institut IAF, Tullastr. 72, 79108 Freiburg, Germany
Thiel & Partner GmbH, Kantstr. 1, 50259 Pulheim, Germany
Keywords: CVD diamond, membrane, loudspeaker, acoustical simulation
Conventional loudspeaker membranes made of metal or synthetic material such as fabric, ceramics
or plastics suffer from nonlinearities and cone breakup modes at fairly low audio frequencies. Due
to their mass, inertia and limited mechanical stability the speaker membranes made of conventional
materials cannot follow the high frequency excitation of the actuating voice-coil. Low sound
velocity causes phase shift and sound pressure losses due to interference of adjacent parts of the
membrane at audible frequencies.
Therefore, loudspeaker engineers are searching for lightweight but extremely rigid materials to
develop speaker membranes whose cone resonances are well above the audible range. With its
extreme hardness, paired with low density and high velocity of sound, diamond is a highly
promising candidate for such applications.
We report on the realization of dome shaped CVD diamond membranes by deposition on curved
silicon substrates. Domes with diameters between 20 and 65 mm and with a thickness ranging from
50 to 120 µm were prepared. After deposition, the substrate is dissolved and the rim of the diamond
dome is cut by laser scribing. Free standing diamond membranes are mounted onto dynamic voice
coils and integrated into tweeter and/or midrange driver chassis. Extended tests and optimisations
led to loudspeaker systems that show a second and third harmonic distortion behaviour in the
important frequency range between 3 to 10 kHz that is reduced by 40% in comparison to already
excellent established values obtained with sapphire membranes. Cone resonance frequencies of
CVD diamond membranes are increased by a factor of two, as predicted by simulations.
Diamond membranes – in theory – have always been the dream of loudspeaker engineers since the
mechanical properties of diamond are close to those of an ideal hypothetic material which would
have an infinite Young’s Modulus and – at the same time – a vanishing density. However, since
nature doesn’t offer diamond in the shape of membranes and the High Pressure High Temperature
synthesis only provides small crystallites, acoustic engineers were compelled to use ordinary
materials like aluminium or sapphire for high end tweeters.
With the development of CVD technologies to produce large area diamond discs with properties
matching those of the best natural diamonds [1,2] and molding techniques to deposit CVD diamond
on preshaped substrates [3,4,5], diamond domes for loudspeaker applications became feasible. After
early suggestions by Sumitomo [6] no further development was done until in 1999 when Fraunhofer
IAF and Thiel & Partner GmbH, an audio equipment manufacturer specialized in hard material
loudspeaker membranes, entered a cooperation aiming to develop a novel high frequency tweeter
with a diamond membrane. This cooperation succeeded in launching its first commercial tweeter
system in the year 2000. Nowadays leading audio system manufacturers such as Avalon Acoustics
(USA), Lumen White (Austria), Mårten Design (Sweden) and Kharma (Netherlands) are using
these tweeters for their high end loudspeaker systems
In this paper we describe the deposition and machining of a diamond membrane, its mounting on a
voice-coil and its integration into tweeter chassis. We will assess the mechanical and acoustical
properties of these systems by FEM simulations and compare their result with sound pressure
measurements. Finally we will discuss the effect of frequencies above the audible range (>20 kHz)
on the auditor.
Preparation of CVD diamond tweeter membranes
To prepare dome shaped diamond membranes, polycrystalline CVD diamond is deposited onto
preshaped substrates. The possibility to make three dimensional diamond devices by simply
replicating a substrate has been used by a number of scientists to create for example anti reflection
structures by the moth eye effect [4] or optical diamond lenses e.g. for CO2-laser surgery [3]. For
these kinds of applications generally silicon is used as a substrate since it is easy to machine and
polish, it enables the nucleation of diamond and is easily dissolved in acid after diamond deposition.
Figure 1 illustrates the processing steps to prepare dome shaped CVD diamond membranes.
Fig.1: Processing steps to prepare dome shaped CVD diamond tweeter membranes. a) preshaped
silicon substrate. b) substrate after deposition of CVD diamond. c) diamond dome after separation
from the substrate. d) laser cutting of diamond dome. e) diamond membrane assembled with
surround and voice coil former.
For growing thick free standing CVD diamond, a microwave plasma CVD reactor with ellipsoidal
cavity [7] that provides stable deposition conditions is used. As standard growth conditions we use a
microwave power of 6 kW at a frequency of 2.45 GHz, a substrate temperature between 700 and
900°C and a pressure ranging from 100 to 200 mbar. The feed gas is 1-2% methane in hydrogen.
After CVD diamond deposition (Fig.1b) the silicon substrate is dissolved in acid (Fig.1c) and the
rim of the now free standing membrane cut using a Nd:YAG laser scriber (Fig.1d). As a last step
the completely machined membrane (Fig.1e) is mounted onto a voice-coil, fixed with a surround
and integrated into a tweeter and/or midrange loudspeaker driver chassis (Fig.2).
Fig.2: Free standing dome shaped CVD diamond membranes (front) and a D20-tweeter from Thiel
& Partner equipped with a CVD diamond membrane.
Mechanical and acoustic properties of diamond tweeters
To describe the acoustic properties of loudspeakers, a commercial finite element modeling program
(Finecone) that calculates amplitude response and impedance in radial symmetry was used. As a
matter of fact, there are also non axial-symmetric modes present, but since these are higher in
frequency than the cone breakup (i.e. the first and dominant natural oscillation of the membrane
voice coil configuration), these do not play a dominant role and therefore we do not want to go in
further detail here. As a first approach and to check the reliability of these simulations, the
resonance frequencies fn of a flat membrane that are given analytically by [8]:
fn = n
Et 3
12 ρ r 4 (1 − ν 2 )
were calculated.
In equation (1) r is the radius and t the thickness of the membrane, E its Young’s Modulus, ν
 its
Poisson’s Ratio and ρ its density. kn is the constant for the n natural oscillation (k1 = 10.2,
k2 = 21.3, k3 = 34.9, …). According to equation (1) the natural frequencies are proportional to the
square root of E/ρ, i.e. a good membrane should be stiff and light. For a flat 60 µm thick diamond
membrane 20 mm in diameter with E = 1140 GPa, ν
 = 0.1 and ρ
 = 3.51 g/cm3, the first natural
frequency occurs according to (1) at a frequency of as low as 5.09 kHz. For a simple rim-supported
membrane the simulated result equals exactly the one analytically postulated. A real tweeter
membrane however is not free standing but attached to a voice-coil former and clamped to a
centralizing surround, both adding mass and increased damping to the whole structure. A cross
section of a typical configuration used for simulations with Finecone is shown in figure 3.
Fig.3: Cross section through a tweeter membrane attached to a former (green) on top of an actuating
voice-coil (red) and clamped to a surround (blue).
For the fixed membrane, the cone breakup occurs at a lower frequency, basically due to the fact that
combined oscillations of former and dome are lower in frequency than oscillations of the isolated
dome itself. Besides the material of the membrane, also its geometry and the properties of the
attached surround and voice coil former play a crucial role for its overall performance. The sag
height of the membrane has a very strong influence on the sound pressure level (SPL). Figure 4
displays the frequency dependence of the calculated SPL for diamond tweeter membranes fixed as
shown in figure 3 for various sag heights.
Fig.4: Sound pressure level (SPL) in a distance of 1m vs. frequency for a 20 mm diamond
membrane with increasing sag height ranging from a flat membrane to a membrane with a sag
height of 4 mm.
Although flat membranes would be highly desirable in terms of vanishing phase loss (see below)
figure 4 clearly states that the sag height has a strong influence on cone breakup frequencies and
high frequency SPL. A comparatively moderate sag height of 0.5 mm increases the frequency of the
first cone breakup almost by a factor of 5 from 3 kHz to 14 kHz with further increase to almost 70
kHz given a sag height of 4 mm. Inversely proportional to the sag height appears the high frequency
SPL and the peak energy of the cone breakup, suggesting an optimum of 3.0 to 3.5 mm sag height.
Cone breakup modes are always associated with a significant increase in harmonic and non-linear
distortion, also with undesired lobing of sound pressure. In terms of sonic quality therefore cone
breakup modes should be pushed as far away as possible from the audible range to avoid any kind
of subharmonic interference.
The membrane thickness has a moderate influence on the SPL, the thicker and thereby stiffer the
membrane, the higher the cone breakup, which is desirable. However for a thicker and heavier
membrane the total SPL is reduced; an optimum can be found for the D20-tweeter in between 50
and 60 µm. Figure 5 compares the SPL of an aluminium, a diamond and a infinitely stiff dome
(∅: 20 mm, h = 3.5 mm) of the same mass. Surprisingly at first sight is the fact that the infinitely
stiff membrane shows a pronounced dip in the response around 90 kHz. This is a result of phase
interferences or phase losses, i.e. for inverted dome shaped membranes matters that the sound
generated at the centre of the membrane reaches the auditor later than the sound generated at the
rim, thus enabling the sound waves to interfere and extinguish each other at certain frequencies.
This effect is strongest if the competing regions have about the same area.
Fig.5: SPL vs. frequency for an aluminium, a diamond and an infinitely stiff membrane, all of the
same mass.
In the case of diamond, the cone breakup at 60 kHz overrides phase losses and as a result prevents
the sound pressure from decreasing. The aluminium dome suffers from cone breakup at a rather low
frequency of 22 kHz although this is well above the audible range. We will discuss in the next
section whether or not this has an influence on the auditor.
Figure 6 finally shows the measured sound pressure level and total harmonic distortion (THD) for
D20-6 tweeter with a diamond membrane. As predicted by theory, the cone breakup occurs at a
frequency of around 65 kHz.
Fig.6: Sound Pressure Level and Total Harmonic Distortion vs. frequency of a D20-6 tweeter,
measured with B&K 4138 microphone in app. 10 cm of distance and R&S UPD Analyzer
One of the most crucial features of the diamond membrane is the fact that the harmonic distortion is
greatly reduced as can be seen from figure 6. In the most important range from 3 kHz to 10 kHz, the
total harmonic distortion in general is below 0.1 % with the exception of occasional spikes. The
mean smoothed value is near 0.07 %, a value we have not been able to achieve with even the best
conventional tweeters. Common THD values of ordinary tweeters are one order of magnitude
Commercially available domes have sizes of 20 mm, 25 mm, 30 mm and 50 mm. Laboratory
samples of 63 mm diameter have been produced. As expected, a similar picture as for the
mentioned 20 mm domes appears with larger cones. Figure 7 shows measured and simulated SPLs
of a 50 mm diamond and sapphire cone. In good agreement with the simulation, the cone breakup
occurs for the diamond dome at about 30 kHz, almost one octave higher than for the sapphire cone
of the same size. In addition the total harmonic distortion (THD) of a diamond cone is way better
than with any kind of conventional membrane material, thus improving the sonic quality of mid
Since the large diamond cone still has a bandwidth that reaches well above the audible range, it can
be used as tweeter with a very low cross-over point. Frequencies as low as 400 Hz appear possible,
thus coming closer to the ideal sound transducer: a single-point, time-coherent minimal-phase
system covering the entire audio band.
Fig.7: Measured and simulated (20 dB offset) SPL vs. frequency for a diamond (solid) and a
sapphire midrange driver (dashed), both 50 mm in diameter.
Membranes made of CVD diamond enable audio frequencies of up to 100 kHz to be transmitted,
values that are not achievable with any other material.
So far, the question whether or not frequencies above 20 kHz have any influence on the auditor had
been of concern only for a few high end audiophiles, who do not use conventional CDs containing
almost no acoustic information above 20 kHz. However the breakthrough for SACD and DVD
audio format with sampling rates of 192 kHz or higher is only a matter of time. Do we all need to
buy new speakers with diamond tweeters?
The perception of frequencies beyond 20 kHz has been lively discussed in the past. It is common
understanding now among audiophiles that extended reproduction chains improve the overall
performance. The degree of frequency extension and the influence on clarity, ambience and
neutrality are still controversially disputed. However various researchers proved that orchestral
music contains significant spectral energy in frequencies up to 100 kHz, Oohashi [9] provided
evidence that inaudible high-frequency sounds have a significant effect on human brains. This has
been called “the hypersonic effect”. Although more research has to be done on the influence of
ultrahigh frequencies on the auditor, the audio community agrees on the fact that systems that are
linear up to highest frequencies sound more detailed, effortless and produce a more realistic sound
stage. This is mirrored by the steadily growing demand for speakers with diamond membranes.
E. Wörner, Thermal Properties and Applications of CVD Diamond, Low Pressure Synthetic
Diamond: Manufacturing and Application, eds. B. Dischler and C. Wild, Springer, Heidelberg,
Germany, (1998) p. 165
C. Wild, CVD Diamond for Optical Windows, Low Pressure Synthetic Diamond:
Manufacturing and Application, eds. B. Dischler and C. Wild, Springer, Heidelberg, Germany
(1998) p. 189
E. Wörner, C. Wild, W. Müller-Sebert and P. Koidl, Diamond Relat. Mater. Vol. 10 (2001) p.
V. Ralchenkov, I. Vlasov, V. Konov, A. Khomich, L. Schirone, G. Sotgiu and A.V. Baranov,
Proc. Applied Diamond Conference / Frontier Carbon Technology´99, eds. M. Yoshikawa et
al., Tsukuba, Japan (1999) p. 128
H. Björkman, P. Rangsten, U. Simu, J. Karlsson, P. Hollman and K. Hjort, MEMS’98,
Heidelberg, Germany (1998) p. 34
Japanese Patent Office: JP59143499A2
M. Füner, C. Wild and P. Koidl, Appl. Phys. Lett. Vol. 72 (1998) p. 1149
W.C. Young: Roark’s Formulas for Stress and Strain (McGraw-Hill Inc., New York, 1975)
T. Oohashi et al, J. Neurophysiol. Vol. 83 (2000), p. 3548