KEF R&D
LS50 Meta
LS50 Wireless II
CONTENTS
CONTENTS CONT’D
1
1
1
1
2
2
4
4
5
6
7
7
9
9
10
11
11
11
12
12
12
13
13
13
13
14
14
14
15
Introduction
Philosophy
LS50 Meta
Tweeter
Metamaterials
Tweeter Metamaterial Absorption Technology
Coupling the Absorber to the Tweeter Dome
Tangerine Waveguide
Tweeter Gap Damper
Motor Design
Bass/Midrange
Motor Design
Crossover
Industrial Design
Specication - LS50 Meta
Appendix 1 - Cabinet
Diffraction
Rigid Bracing/Constrained Layer Damping
Appendix 2 - Ports
Resonances inside the cabinet
Organ pipe resonances
Turbulence
Appendix 3 - Uni-Q
Tweeter
Diaphragm
Tangerine Waveguide
Bass/Midrange driver
Diaphragm
Z-ex Surround
17
18
18
18
19
20
21
20
21
21
23
23
24
24
24
25
26
LS50 Wireless II
Overview
Inputs
Connection between Primary and Secondary
Streaming support
DSP Processing
EQ Settings
Desk & Wall Modes
Treble Trim
Phase correction
Bass Extension
Adding a Subwoofer
Syncing with Vision
Firmware
Power Amplication
Summary
Specication - LS50 Wireless II
16
16
Summary
References
Introduction
“If I have seen further, it is because I stand on the shoulders
of Giants.”
This famous quotation, attributed to Sir Isaac Newton,
illustrates how progress is made in virtually all
scientic disciplines - progress is evolutionary, rather
than revolutionary. So it is in the development of
speakers and techniques developed for one model are
carried through to subsequent designs and added to.
That small loudspeakers could be serious hi-
reproducers was proved in the 1960s with Laurie
Fincham’s Maxim design for Goodmans. Grossly
inefcient, with a cone tweeter and rudimentary
crossover, this speaker was outdated by the 1970s,
but nevertheless inspired the BBC to design the
larger, but still diminutive LS3/5. Originally designed
for ¼ scale auditorium acoustic modelling, this design
was soon recognised for its suitability as a monitor in
the cramped surroundings of BBC Outside Broadcast
vans. The design was subtly changed to become the
legendary LS3/5A when the KEF drive units, with
which the speaker was equipped, were modied.
As time went by and loudspeaker design progressed,
it was again time to review the capabilities of the
LS3/5A and in 2012, to mark the 50th anniversary of
KEF, the original LS50 was born. As the name suggests,
the LS50 Meta is a development of the original LS50,
and again redenes what a small monitor loudspeaker
can do.
These words set the company’s philosophy from
day one and they are as relevant today as they have
ever been. Continuing research into materials and
engineering has always driven KEF’s quest for
innovation, with the added element of ensuring that
the resulting designs work within a typical domestic
environment. Everybody deserves great sound.
LS50 Meta
Both LS50 Meta and LS50 Wireless II feature a 12th
Generation Uni-Q driver array featuring an industry-
rst - Metamaterial Absorption Technology (MAT).
This particular Uni-Q driver, whilst being inspired by
that of the R Series, has been engineered with regards
to the overall system design.
Firstly, the addition of MAT has necessitated
considerable redesign of the driver structure, allowing
KEF engineers to further rene technologies such as
the tweeter gap damper.
Secondly, where R Series features 3-way designs, LS50
Meta is a 2-way loudspeaker, meaning the midrange
driver must cover a much wider bandwidth, down into
bass frequencies. Whilst excellent for integration, this
does provide challenges in regards to performance.
With this in mind, KEF engineers made a number of
changes to the motor system, suspension, surround
and cone of the midrange driver to deliver as smooth a
performance as possible.
Tweeter
LS50 Meta - Carbon Black
Figure 1
Philosophy
“Of all art, music is the most indenable and the
most expressive, the most insubstantial and the
most immediate, the most transitory and the most
imperishable. Transformed to a dance of electrons
along a wire, its ghost lives on. When KEF returns
music to its rightful habituation, your ears and mind,
they aim to do so in the most natural way they can…
without drama, without exaggeration, without artice.”
Raymond Cooke, KEF founder
With any drive unit, as much sound is generated at the
rear of the unit as at the front (see gure 2) and this
radiation is unwanted and needs to be absorbed.
Figure 2 - Generic driver
1
The most important innovation in this loudspeaker
is the near-perfect absorption of the unwanted rear
sound generated by the tweeter dome. One might
legitimately ask, “Why is this important?”
It’s true that, if the tweeter is listened to on its own
without the contribution of the bass/midrange driver
- and here we are talking of frequencies above about
3kHz - very little seems to come out of it. More
revealing is to listen to any loudspeaker system
without the tweeter connected. The sound is mufed
and, although one can discern that it’s a piano or a
violin playing, it’s impossible to hear the difference
between a Steinway and a Bosendorfer, a Stradivarius
and an also-ran violin.
The true music lover or audiophile wants to hear this
amount of detail. More than that, one should be able
to imagine the striking of the keys or the way the bow
is stroking the strings. In short, it should be the closest
one can get to a live performance without actually
turning up.
the mathematical expression between the target
absorption spectrum and the physical realisation,
which has a minimum thickness only 1/10th of the
wavelength at the lowest frequency (620Hz in this
case).
Tweeter Metamaterial
Absorption Technolology
Figures 3 and 4 respectively show a computer model
of the absorber and the real thing. Its design borrows
from a room acoustics absorber in that it is based
around ¼ wavelength resonators. Looking rather
like a maze, what you can see are 15 separate tubes,
folded up to make a circular object. In fact, there are 2
layers, so there are, in total, 30 tubes, each tuned to a
different frequency.
Each tube has a very high Q, which means high
absorption over a very narrow band, and the only
damping, which is the way the energy is dissipated,
comes from the friction between the air moving in
the tubes and the tube walls. The frequency spacing is
designed so that, when the effects of all the tubes are
combined together, the absorption is uniform over a
wide band.
Figure 5 shows the pressure response, measured at
the closed end of each of the 30 channels, together
with the absorption spectrum of the whole absorber.
This sort of detail comes from the tweeter and
the better the tweeter, the more the enjoyment of
listening to the music.
Metamaterials
Metamaterials are probably most familiar in the eld
of optics, where synthetic materials may be realised
that have properties that cannot be found in nature.
For example, a base material may be infused with
another in varying density such that the refractive
index varies throughout the material. Thus things like
at lenses may be constructed that are much easier to
produce than grinding glass to a precise shape.
The term “meta” has since gained the more general
description of any material that exhibits characteristics
foreign to the solid form. In this case, ABS has been
moulded into a shape that is an almost ideal broadband absorber.
The design of the absorber was a joint project between
KEF and AMG (Acoustic Metamaterials Group).
With the help of Professor Ping Sheng (a worldrenowned expert on metamaterials), AMG derived
Figure 3
Computer model of absorber
Figure 4
Actual absorber
Top: pressure response at closed end of each separate absorber channel.
Bottom: absorption spectrum of whole absorber.
Figure 5
Figure 7
Top: Amplitude and Phase responses of transfer functions
V1 (green), V2 (blue) and V1+V2 (red).
Bottom: Amplitude and Phase responses of input impedance
These are steady-state measurements and anyone
who is familiar with lters will know that, although
the amplitude response of narrow-band lters may
well add to unity, this usually hides the presence of allpass phase responses and some time smearing occurs.
The system is said to be non-minimum phase. This is
the case with the summing of the transfer functions
of the separate channels of graphic equalisers and
the outputs of two or more drivers fed by a passive
speaker crossover network. Such time smearing can
readily be shown by impulse measurements, where a
Generalised 3rd-order Butterworth constant-resistance
Figure 6
crossover. Units are Ohms, Farads and Henries. fc is cut-off
2 3
change in prole between the exciting signal and that
generated by the device(s) indicates time smearing.
Coupling the Absorber to the
Tweeter Dome
The absorber sits at the rear of the Uni-Q™ driver
and is coupled to the tweeter dome by a slightly
tapered conical duct, which acts as a waveguide. This
waveguide passes through the centre poles of both the
tweeter and bass/midrange drivers and has involved a
complete redesign of the tweeter magnet assembly
to accommodate the wider diameter required for the
duct to work properly. The difference in the motor
assemblies is shown in gure 9.
not being totally absorbed. It exposed a lesser, but still
important layer of residual colouration. It turned out
that the tangerine waveguide and the dome surround
support - both plastic mouldings - were physically
deforming at high frequencies. (gure 11)
Tweeter Gap Damper
One of the problems with constructing a combination
driver array like Uni-Q is dealing with the gaps that
separate the constituent parts. There is a narrow
channel – an annular gap – between the moving
midrange voice coil and the static start of the tweeter
waveguide. This channel acts as an organ-pipe-like
resonator, and is excited by the tweeter output. The
resonances modify the response of the tweeter, adding
a series of glitches that are not present if the gap is
closed off – simulating a perfectly smooth waveguide.
Impulse response of tweeter dome velocity (red), overlaid with
Figure 8
ideal response (dotted blue)
However, it must be realised that what is being used
here is the acoustic impedance, not a transfer function.
To take an electrical analogy, consider a passive,
constant-resistance network such as that of gure 6.
While the individual low- (V1) and high-pass (V2)
transfer functions are minimum phase, the sum of the
transfer functions (V1+V2) is not. It is a 2nd-order allpass. However, the total input impedance (Zin), is a
pure resistance, R. As such, it is minimum phase. There
is no all-pass (gure 7).
To illustrate that the absorber is indeed minimum-
phase and introduces no time smearing, gure 8 shows
the impulse response of the tweeter dome movement
overlaid with an ideal, minimum phase response.
There is virtually no difference.
The tweeter dome movement was chosen because it
can be measured close-to, which increases the signal
to noise ratio. As it is only valid well below the rst
break-up mode of the dome and the absorber only
works above a certain frequency, both signals were
equally band limited to avoid errors.
A small amount of porous material is placed in the duct,
which has the dual effects of reducing the amount of
ripple at high frequencies and ne-tuning the knee
of the absorption spectrum. Figure 10 shows the
absorption spectrum immediately behind the dome.
Figure 9
LS50 (left) and LS50 Meta (right) tweeter motor assemblies.
Figure 10
Absorption at the entrance of the conical duct, immediately
behind the dome diaphragm.
Tangerine Waveguide
Many acoustic engineers would be so pleased at
designing the almost perfect tweeter absorber that
they would have rested on their laurels. But speaker
design is rather like peeling an onion - remove one
layer and there is another one exposed. So it is with the
removal of colouration caused by the rear radiation
FEA simulation of exaggerated deformation of tangerine
Figure 11
waveguide and surround support at 12kHz
Strengthening ribs were added to both components,
which reduced the deformation. Figure 12 shows
the actual modied parts (viewed from the rear) and
gure 13 illustrates the reduction in displacement.
It should be noted that the modied parts have an
area comparable to the tweeter dome itself and any
movement will add audible colouration to the overall.
For the basic operation of the tangerine waveguide,
featured in previous models, see Appendix 3.
Figure 12
Modied surround support and tangerine waveguide viewed
from the rear to show added ribs
Figure 13
Simulated displacement of tangerine waveguide
Obviously, this annular gap is necessary to allow the
LF/MF cone and voice coil to move, so the solution was
to create a cavity between the midrange and tweeter
magnets to which this annular gap connected (gure
14). Adding damping to this newly-created cavity was
found to be effective in taming the resonances in the
annular gap and the removal of the response glitches
was immediately apparent as an improvement in
detail clarity (gure 15).
This feature - the tweeter gap damper - was rst
introduced through the R Series (2018), and was
a major development that led to birth of the 12th
Generation Uni-Q. The addition of MAT and the
Figure 14
Uni-Q driver with damped, optimised gap
change in tweeter system structure for LS50 Meta,
however, necessitated a redesign of the tweeter gap
damper. Additional work was carried out in regards
to the shape of the cavity and the placement of the
wadding material - now comprising of two rings - to
further improve performance.
4 5
the back of the dome is directed into the metamaterial
absorber. but also reduces non-linear distortions
caused by air pressure changes behind the tweeter.
As a tweeter moves in and out, it compresses and
decompresses the air behind it. This negatively
affects the excursion characteristics and linearity of
the tweeter The larger the air volume, the lesser the
effect.
Also notable is the increased saturation of the steel
components of the motor system. The image colour
is indicative of the magnetic strength throughout the
steel, and dark blue indicates saturation.
Bass/Midrange
Motor Design
The design of the motor used in the bass/midrange
driver of the Uni-Q driver has also been modied to
reduce modulation by the movement of the voice
coil, but this time the approach is different from that
used for the tweeter, partly because the driver has an
overhung voice coil (coil longer than the magnet gap)
as opposed to the underhung coil (coil shorter than
the magnet gap) of the tweeter, which is always fully
in the steel gap.
Figure 15
Tweeter response
With damped cavity (blue)
Without cavity (red)
Motor Design
In addition to the tweeter motor system being
modied to accommodate the wider duct through its
centre pole, other changes were made to provide a
more consistent drive to the voice coil.
Figure 16 illustrates the difference in tweeter motor
design between the original LS50 and the LS50 Meta.
Figure 16
Tweeter motor systems of LS50 (top)
and LS50 Meta (bottom).
One major change is the sheer size difference of
the duct - 2.5x the surface area of that found in the
original LS50, increasing the air volume available to
the tweeter.
This ensures as much of the sound as possible from
Normally the magnetic eld in the gap is perturbed by
the electric current in the voice coil, which changes
the force on the voice coil during every cycle of
movement. Perturbation cannot occur if the motor
system is saturated and so the force on and therefore
the motion of the voice coil more closely follows the
input signal.
Saturation did result in a drop of ux density in the gap.
This was compensated by changing the tweeter voice
coil from one to two layers, which would normally lead
to an increase in coil inductance, with a progressive
loss of sensitivity at very high frequencies. A copper
sleeve on the centre pole, which couples inductively
to the voice coil, reduces the inductance, even below
that of the LS50, as is shown by the blocked coil
measurements of gure 17.
Figure 17
Tweeter blocked impedance
The inductance of a voice coil presents a couple of
challenges. Higher levels of inductance create a larger
net magnetic eld as the signal passes through the
voice coil. This magnetic eld causes distortion - the
lower the inductance of the voice coil, the lower the
net magnetic eld is produced, thus lower distortion.
The inductance of the voice coil also tends to change
as the coil moves in the gap (modulation). The steel
acts as a core to increase the coil’s inductance and
the amount of steel (and therefore the inductance)
varies as the coil position changes. This variance is an
issue, as it introduces non-linear distortions into the
loudspeaker performance.
As a result, more extensive use is made of auxiliary
aluminium parts in LS50 Meta than was the case
with the original LS50. The aluminium, although non-
magnetic, is conductive and couples inductively to the
voice coil. Being a virtual short circuit, it reduces the
value of the coil’s inductance, which in turn reduces
modulation of the ux of the motor system - reducing
non-linear distortions.
The LS50 and LS50 Meta motor designs are compared
in gure 18. The aluminium parts are coloured light
grey.
Where the original LS50 features a small aluminium
ring placed between the magnet and voice coil,
LS50 Meta sports a much larger version. In addition,
another ring has been added to the top. The red
fringing in the otherwise blue steel is a measure of the
coil modulation and is seen to be much lower in the
LS50 Meta.
Bass.midrange motor systems of LS50 (top)
Figure 18
and LS50 Meta (bottom).
The LS50 Meta motor system also features an
undercut pole design to minimise the steel volume
and focus the magnetic eld more precisely on the
voice coil. Also instructive is to measure the voice
coil inductance at different frequencies and compare
different designs.
For the measurements in gure 19, the voice coil is
blocked (glued in place) at the centre (rest) position.
LS50 Meta - with the shorting rings - exhibits much
lower inductance than the other designs.
Inductance modulation was also measured at two
different frequencies (200Hz and 2kHz) against
excursion. Figure 20 clearly shows that the LS50
Meta design, with the shorting rings, exhibits the least
variation of inductance against excursion - creating
a frequency response that stays balanced, with very
little inuence from the excursion of the driver.
6 7
Crossover
The topology of the crossover of the LS50 Meta is
shown in gure 24. The smoother response of the
new tweeter has resulted in fewer components in the
HF lter and the remaining series capacitor (C1) is of
higher quality.
the HF lter is on a separate board, placed well away
from the LF lter.
Bass/midrange blocked voice coil inductance against frequency.
Figure 19
The change in magnet design has had two further
advantages. The undercut centre pole gives higher
ux density in the gap itself and lower magnetic
fringing (gure 21). This results in higher sensitivity of
the driver and more linear drive at higher excursions
(gure 22).
Looking at gures 18-22 together, we see that the
new design gives less modulation of the magnet ux
density in the gap, more linear drive, lower voice coil
inductance and virtually no change in inductance with
voice coil position. The upshot of all this analysis and
redesign is a signicant reduction in total harmonic
distortion (THD) in the frequency range where the ear
is at its most sensitive (gure 23).
Bass/midrange voice coil inductance against displacement
Figure 20b
at 200Hz (gure 20a) and 2kHz (gure 20b).
Figure 21
Comparison of ux density in the gap.
Figure 24
LS50 Meta crossover topology
The LF lter is similar to that in the original LS50, but
includes an extra low-Q parallel resonance branch (C3R3-L4) to compensate for the higher sensitivity in the
driver’s upper midrange due to its lower inductance.
Current consumption of inductors L2 and L4
Figure 25
for an input of 2.83V.
Industrial Design
Little has changed from the original LS50, but there
are a few detailed changes to the back of the cabinet
to improve performance and enhance looks (gures 1
& 26):
• The port opening is now ush with the rear
surface, which further reduces turbulence.
• The 4 cavities in the corners for the bafe
retention bolts have been eliminated.
• The rear surface is now slightly domed.
• A step transition between the rear panel and
the sides of the cabinet has been introduced.
Figure 20a
Comparison of bass/midrange force factor
Figure 22
Figure 23
THD, measured as a percentage of the fundamental level
90dB SPL at 1m for LS50 (red) and LS50 Meta (blue).
The inductor L4 is the only one in the crossover to
have a core. It is vital that only the resistor R3 controls
the Q of the branch. To that end, the resistance of
• There is now a soft notch around each
binding post.
L4 (which is subject to change as it heats up when
passing current) is kept as low as possible. As the core
is a possible source of harmonic distortion if it nears
saturation, measurements were taken to relate the
current passing through L4 with that passing through
L2. Figure 25 shows that it will be seen that the current
passing through L4 is much lower and the chances of
its core saturating are virtually nil, a fact borne out by
the THD measurements of gure 23.
To minimise any coupling between inductors, the
three in the LF lter are arranged orthogonally and
8 9
Rear views of LS50 (left) and LS50 Meta (right)
Figure 26
Specication - LS50 Meta
Description
Drive Units
Frequency Range (-6dB) 47Hz - 45kHz
Frequency Response (±3dB) 79Hz - 28kHz
Sensitivity (2.83V, 1m) 85dB SPL
Maximum SPL at 1m 106dB
Harmonic distortion <0.4% 175Hz - 20kHz (90dB SPL, 1m)
Crossover frequency 2.1kHz
Nominal Impedance 8Ω
Minimum Impedance 3.5Ω
Dimensions (inc. terminals) H: 302mm (11.9 in)
Weight 7.2kg (15.8 lb)
Written specications like these, however useful for comparison, do not fully describe the listening experience.
However good the component parts, voicing the system with the crossover is all-important.
The following graphs, which include additional directivity and early reection information in line with the
research work of Floyd Toole [9] [10], show how smooth the response of the LS50 Meta is, indicating a superb
listening experience.
Bookshelf Loudspeakers
Uni-Q™ Driver Array
LF/MF: Nominal dia. 130mm (5.25 in) magnesium/aluminium alloy cone
HF: Nominal dia. 25mm (1 in) aluminium alloy dome vented with
Metamaterial Absorption Technology (MAT)
W: 200mm (7.8 in)
D: 278mm (10.9 in)
Appendix 1 - Cabinet
The cabinet construction follows closely that of the
original LS50.
Diffraction
Diffraction is the secondary radiation that happens
when the wavefront from the driver meets a sharp
edge. This reradiation causes what might be described
as a mini-echo that time-smears the driver’s output.
The effect varies with angle, but needs to be minimised
to maintain clarity.
At very low frequencies, rather like a sea wave
wrapping round a small pebble, the sharp edges of
the cabinet have little effect - the sound simply wraps
around the object in a smooth manner. At very high
frequencies, only a minimal proportion of the energy
travels along the front bafe, due to the increased
directivity of the radiating diaphragm, and the
reradiation level is low. In between these extremes,
the shape of the cabinet, especially the front bafe,
has a marked effect.
Traditionally, it was thought that the edges of the bafe
need to be rounded, preferably with a radius similar
to the wavelength of the sound. However, detailed
research has shown that, if the bafe is curved, the
angle at the edge does not have to reach 90 degrees.
for the majority of the diffraction effect to be
eliminated. This is extremely fortunate, because fully
rounding the corners with a large radius is wasteful of
space and not very nice to look at. The front bafe is
curved in both directions enough to make diffraction
inaudible.
Rigid Bracing/Constrained
Layer Damping
The ultimate goal in the design of any speaker is to
have all the sound energy coming from the drivers
and none from anything else. The cabinet walls are a
particular problem as they have a much larger area
than the driver diaphragms and their output can be
problematic with relatively little exure. Not only is
the sound delayed relative to the direct sound from
the drivers, but it’s highly distorted and the resulting
colouration can be extremely unpleasant.
Vibration of the cabinet walls is excited both by
variation in air pressure inside the cabinet and by the
reaction force on the driver motor as the diaphragm
moves back and forth. This latter force is transmitted
via the driver chassis to the cabinet panel.
That the cabinet needs to be there at all may be
realised by going back to gure 2. The bass/midrange
driver radiates energy to the rear and that energy
has to be contained, otherwise the rear radiation
will progressively cancel the front radiation as the
frequency decreases, and absorbed - which effectively
means turning it into heat.
This constraint has traditionally been done by adding
bracing. On its own, this has less effect than may be
expected from simply considering static stiffness.
What it tends to do is shift the problem to higher
frequencies. However, the simple expedient of adding
a lossy layer between the braces and the cabinet
walls absorbs far more energy - Constrained Layer
Damping. This lossy material is sandwiched between
the bracing and the driver/panel, converting vibration
into heat.
Figure 27
Low-diffraction Bafe
10 11
Constrained-layer damped bracing
Figure 28
Appendix 2 - Ports
Ports are used to augment bass response. They have
the dual advantage that they enable a lower cut-off
frequency than closed-box systems for the same
size of enclosure and, within the operating range of
the port, the bass driver moves less, thus lowering
distortion.
Ports are not without their drawbacks, however.
• They can be a window to resonances in the air
cavity inside the cabinet.
• There is more freedom on the rear panel
compared to the front in positioning the port so it
is placed close to nodes (nulls) of the resonances,
so less energy is transmitted through the port.
To illustrate this last point, gure 29 shows the
difference in output when the port is placed at an
antinode and at a node of an internal resonance. For
the rst of these measurements there is no wadding
inside the cabinet, so the effect is made clearer. The
resonance can clearly be seen in the total response.
When the ports are optimally placed and the wadding
added, the resonances are greatly depressed and
cannot be detected in the overall response.
Organ pipe resonances
These resonances within the port itself can be
drastically reduced if the walls of the port have
a degree of exibility. Instead of the normal rigid
plastic, the walls are fabricated from closed-cell
foam. Originally developed for the original LS50
loudspeaker, this technique dramatically reduces the
pressure variations inherent in these higher frequency
resonances and renders them less audible.
Figure 30 illustrates this technique using Finite
Element Analysis (FEA), where the colour scale from
green through to red indicates the level of air pressure.
Figure 30
Comparison of the Finite Element modelled pressure
magnitude in the port at the rst standing wave with rigid
walls (top) and exible walls (bottom).
Turbulence
This occurs at both ends of the port as the air exits the
port either to the open air or the air enclosed within
the cabinet. The solution is to are both ends of the
port (gure 31). The progressive expansion of the air
afforded by the aring reduces turbulence and thus
audible chufng. It should be noted that, if turbulence
is allowed to develop, not only is it audible, but the
performance of the port changes with sound level,
impairing the dynamic range of the loudspeaker.
Figure 31 - Flow pressure contour of straight port (top)
and ared port (bottom).
Figure 32 shows the nal design of the port located
inside the LS50 Meta cabinet.
Appendix 3 - Uni-Q™
The Uni-Q array has been the mainstay of virtually
all KEF loudspeakers since its introduction in 1988.
It delivers the holy grail of loudspeaker design in that
all sound appears to emanate from the same point
in space. Coaxial loudspeakers had been around for
many years before the introduction of Uni-Q, but the
tweeter was never time aligned with the midrange or
bass/midrange driver it was partnered with. Either
the tweeter was in front of the larger driver, which
brought the added disadvantage that the tweeter
impaired the response of the driver it was in front of,
or it was well behind. With Uni-Q, both drivers have
their acoustic centre at the same point and the larger
driver acts as a waveguide for the tweeter. The result
is that the blend between the two units is virtually
seamless in terms of both response and dispersion.
The two units together can be regarded as a single
driver without the performance shortfall that would
be suffered by a true single driver covering such a
wide frequency range.
Over the intervening years, the Uni-Q concept has
been progressively rened. The midrange cone shape
has been optimised to create just the right amount
of dispersion at all frequencies and innovations
such as the tangerine waveguide have improved
the performance of the tweeter. Uni-Q is simple in
concept, but tricky to implement in practice. Here
are some of the techniques previously developed and
carried over to this model.
• At midrange frequencies, they act rather like an
organ pipe and exhibit a series of resonances.
• They can be a source of noise if turbulence is
allowed to happen as air vents out at each end of
the port (so-called chufng).
Resonances inside the cabinet
The port is positioned on the rear panel of the cabinet.
This has two advantages:
• Resonances inside the cabinet are at relatively
high frequencies and the sounds are not directed
towards the listener.
Tweeter
Diaphragm
The dome diaphragm itself is aluminium. More exotic
materials – diamond and beryllium for example – are
sometimes used in an effort to increase stiffness and
push the pistonic region of the tweeter to the limits of
human hearing. But this can be done with aluminium
at far lower cost, providing some ingenuity is used in
designing how the dome is constructed.
Figure 29
Simulated comparison of the port and total speaker outputs
with unoptimised (top) and optimised (bottom) port location.
Figure 32 - Offset exible port located inside the LS50 Meta
12 13
The optimum dome shape for waveguide loading
is a spherical cross-section, but the optimal shape
for stiffness is an elliptical cross-section. Both were
combined into the patented KEF Stiffened Dome. A
one-piece elliptical dome and voice coil former is deep
drawn and the centre removed. This is capped by a
spherical dome and a very stiff triangular section is
formed where the two parts join.
FEA Simulation of tweeter output
Figure 34
without tangerine waveguide (red)
with tangerine waveguide (blue)
Being within the human audio range, the breakup has to be tamed, rather than putting it out of
hearing range. This is done by introducing a exible,
decoupling material between the cone neck and the
voice coil former.
The high-Q resonances are tamed to the extent that
they may be properly attenuated by the crossover
and not break through the tweeter output. Figure
37 illustrates the improvement in the unit’s response
as a result of using the lossy interface. Mechanical
correction is far better than equalising the peaks with
the crossover. The latter does not subdue peaks in the
THD at sub-multiples of the fundamental.
It will be seen that the outer vertical wall of the
surround is another possible source of discontinuity,
so the outer trim ring is part of the whole design,
giving a smooth transition all the way from the tweeter
diaphragm to the front bafe.
3D CAD sectional views of the tweeter dome and extended
Figure 33
former that meet to form a triangular stiffening member
at the dome edge
Tangerine Waveguide
The interface between the radiating dome of the
tweeter and the waveguide formed by the midrange
dome is extremely critical. Ideally, the diaphragm
should be a pulsating dome, which would involve
the radius of curvature changing. This is not possible
and the motion is in the same direction all over the
surface of the dome. To compensate for this nonideal situation and taking a leaf out of the design of
compression drivers, the Tangerine Waveguide was
developed to restore correct coupling between the
dome and the whole waveguide.
The improved coupling at high frequencies above
5kHz brings with it a useful increase in sensitivity and
a reduction in the height of the rst resonance peak,
illustrated by gure 34.
Figure 35 shows the tangerine waveguide in place in
the Uni-Q array. Note that this is the basic operation
of the tangerine waveguide that is carried over from
previous models. It is enhanced by the extra stiffening
described in the new features section (gure 12).
Tangerine waveguide in position on top of tweeter dome
Figure 35
Bass/Midrange driver
Diaphragm
The cone is formed from a magnesium/aluminium
alloy. Like the tweeter dome, it serves to provide the
necessary stiffness to give pure pistonic motion over
the driver’s working range. The stiffness is increased by
the radial embossing in the cone prole. Nevertheless
the cone is still prone to high-Q break-up, but this time
in the frequency range covered by the tweeter.
Cone Neck Control interface between voice coil former and driver cone.
Figure 36
Z-Flex Surround
The midrange cone and outer surround form a
waveguide for the tweeter and the normal roll
surround is not ideal in this respect, especially close to
the driver’s axis, where symmetry has a large effect.
The Z-Flex surround has been specially designed to
allow the necessary movement of the bass/midrange
cone, but provide a low prole to allow the cone and
surround to act together more accurately as a tweeter
waveguide.
Effect of lossy interface on response of LF/MF driver
Figure 37
Figure 38
Z-Flex surround - line drawing
Figure 39
Cross section of complete Uni-Q driver and trim ring
14 15
References
[1] M. Yang, S. Chen, C. Fu and P. Sheng, “Optimal
sound-absorbing structures,” Mater. Horiz., vol. 4, pp.
673-680, 2017.
FEA simulated tweeter impulse response with half-roll surround (top)
Figure 40
and Z-Flex surround (bottom).
Note the improved response, especially close to the unit’s axis
[2] S. Degraeve and J. Oclee-Brown, “Metamaterial
Absorber for Loudspeaker Enclosures,” 148th Audio
Eng. Soc. Convention, 2020.
[3] KEF Research and Development, “R Series White
Paper,” 2018.
[4] KEF Research and Development, “The Reference
White Paper,” 2013.
[5] M. Dodd and J. Oclee-Brown, “New methodology
for the Acoustic Design of Compression Driver Phase
Plugs with Concentric Annular Channels,” J. Audio
Eng. Soc., vol. 57, no. 10, pp. 771-787, Oct 2009.
[6] M. Dodd and J. Oclee-Brown, “A New Methodology
for the Acoustic Design of Compression Driver Phase
Plugs with Radial Channels,” 125th Audio Eng. Soc.
Convention, 2008.
LS50 Wireless II
Summary
As can be deduced from reading this paper, there is
a lot of technology packed into the LS50 Meta, both
new and existing - a culmination of many years of KEF
innovation and acoustic expertise.
All this technology, although tackling legitimate
sound reproduction problems in a scientically
logical manner, is of little use if the listener is not
brought ever closer to sound Utopia - getting more
and more enjoyment and fullment from the listening
experience.
This is probably the best compact passive monitor
that is available today. This is a bold claim, but valid
not just because of the engineering expertise packed
into it, but also because the engineers responsible for
its design, often frustrated musicians, are fully aware
of what sounds good, what sounds bad and what
shortcomings the ear is able to gloss over.
[7] M. Dodd, “The development of the KEF LS50, a
compact two way loudspeaker system,” International
Conference, Helsinki, Finland, 2013.
[8] A. Salvatti, A. Devantier and D. J. Button,
“Maximizing Performance from Loudspeaker Ports,” J.
Audio Eng. Soc., vol. 50, no. 1/2, pp. 19-45, Jan. 2002.
[9] F. E. Toole, “Loudspeaker Measurements and Their
Relationship to Listener Preferences: Part I,” J. Audio
Eng. Soc., vol. 34, no. 4, pp. 227-235, Apr. 1986.
[10] F. E. Toole, Sound Reproduction, The Acoustics
and Psychoacoustics of Loudspeakers and Rooms,
Amsterdam; Boston: Elsevier, 2008.
Figure 41
Primary Speaker Rear Panel
Figure 42
Secondary Speaker Rear Pane
16 17
Overview
4. Analogue stereo - Aux 3.5mm stereo jack socket
Streaming support
The following music streaming services are also
supported (dependent on territory):
The active LS50 Wireless II incorporates all the
technology described for the passive LS50 Meta, plus
a few extras that either add to the presentation of the
music - making it even more accurate - or add to the
functionality.
The electronics, mounted on the back, includes two
channels of power amplication (one channel for each
driver and each of higher power than the original LS50
Wireless) and the KEF Musical Integrity Engine (MIE).
This is essentially custom Digital Signal Processing
(DSP) that allows the acoustic engineer to do far more
than is possible in any passive design.
Each stereo pair comprises a Primary and a Secondary
speaker. The former incorporates adjustments
applicable to both speakers, such as bass extension,
room acoustic compensation, source selection and
placement/EQ.
5. Network - Wireless (2.4GHz or 5GHz) and Wired
(RJ-45 socket and CAT 6 cable) supporting up to
384kHz/24bit, DSD256, MQA
6. Bluetooth 4.2
Selection is made using the smartphone app, the
controls on the top of the Primary speaker or the
supplied remote control.
Wireless streaming supports the following:
AirPlay 2
Google Chromecast
ROON Ready (via future rmware update)
UPnP compatible
Bluetooth 4.2
Spotify (via Spotify Connect)
Tidal
Qobuz
Deezer
Amazon Music
QQ Music via QPlay
Internet Radio
Podcast
Bass may be extended to lower frequencies than is
limited by the usual passive cabinet-size/sensitivity
trade-off, the balance may be adjusted for speaker
position and room acoustics and the all-pass delay
inherent in passive crossovers may be eliminated.
All adjustments are made using a smartphone app.
This is available for both iOS and Android devices and
is a free download either from the Apple App Store or
the Google Play Store as appropriate.
Inputs
All inputs are fed to the Primary speaker (gure 43).
The six choices available are:
1. eARC PCM at up to 192kHz/24 bit - HDMI (TV)
socket
2. S/PDIF at up to 96kHz/24 bit - TOSLINK Optical
socket
3. S/PDIF at up to 192kHz/24 bit - RCA Phono (coax)
socket
Figure 43
Primary Speaker Inputs
Connection between Primary
and Secondary
The two speakers may be connected together either
wirelessly or hard-wired.
LS50 Wireless II is pre-paired at the factory for easy
setup in wireless mode. The wireless transmission
is totally independent of the local Wi-Fi signal. The
miniscule delay (2.9ms) between the primary and
secondary speakers in this mode is automtically
accounted for, through an equal delay applied to the
primary speaker. Finally, the wireless connection
resamples all audio to 96kHz/24 bit.
The wired interspeaker connection uses CAT-6 cable,
plugged into the relevant sockets on the rear of the
speakers. Cable mode must also be enabled in the KEF
Connect app settings menu. In this mode, all audio
signals are resampled to 192kHz/24 bit.
Figure 44 - Block diagram of Primary Electronics
Figure 45 - Block diagram of Secondary Electronics
Figure 46 - DSP Details
18 19
DSP Processing
Phase correction
Figures 44 & 45 show block diagrams of the electronics
in the Primary and Secondary speakers respectively.
All settings are controlled from the smartphone app
and applied equally to both speakers.
DSP allows much ner control of the signals reaching
the drivers than with a passive crossover, and the
LS50 Wireless II crossover has been redesigned from
the passive version to take advantage of this. Bass
cut-off, although still 4thorder, has a less rounded
corner because there is more level adjustment. By
themselves, these lters still result in an all-pass
response. However, there is no interaction between
the crossover and the driver impedance in the case of
active crossovers, leading to better consistency from
sample to sample. The all-pass characteristic may be
eliminated by the Phase Correction stage and this will
be discussed in greater detail in a subsequent section
of the paper.
EQ Settings
The KEF Connect app allows the listener to make
EQ adjustments (Figure 47) based on the listening
environment and speaker position.
EQ settings are available in both Basic and Expert
modes. Both feature largely the same settings, with
the main difference being the way parameters are
presented - generally distances for Basic and decibels
for Expert.
Recommended wall mode settings
Figure 48
Figure 49
Recommended desk mode settings
As stated before, the basic crossover introduces an
allpass into the speaker’s response. This does not
impinge the magnitude response, which may still
be smooth and at, nor does it affect the directivity,
but it does introduce some phase distortion with its
accompanying time delay.
However, it is possible, with an active crossover, to
compensate for this phase distortion and, if the Basic
mode setup is used, this phase compensation is applied
by default; the toggle switch is only provided in Expert
mode. This enables listeners to hear the difference which is subtle, but there. There is also a small amout
of delay involved with the correction. This extra delay
is unlikely to be a problem (see the later section Syncing with Vision), but the overall latency can be
reduced if necessary by switching the correction off.
At this point, it is worth emphasising just how well the
Uni-Q principle lends itself to this type of correction better even than physically separated drivers that are
staggered to correct for time delay on axis.
Here is a conventional speaker with the drivers
mounted on a stepped bafe so that the distances to
the measuring microphone on axis (red for the bass
driver, black for the tweeter) are equal.
Desk & Wall Modes
These two modes are used in conjunction to equalise
for the method of mounting. Below the on/off toggle
is a slider showing the selected value. The line below
the slider shows the applicable range. However, the
mechanism of nearby reections is very complicated
and simple level settings are only an approximate
solution. We would always recommend stand
mounting, well away from walls for optimum results.
Figures 48-50 show some recommended settings.
Expert EQ Settings page (KEF Connect)
Figure 47
Recommended combined wall & desk mode settings
Figure 50
Treble Trim
This is used to compensate an over-lively (lots of glass/
at surfaces) or over-dead (lots of soft furnishings,
Conventional speaker, mic. on axis.
curtains) room acoustic. The former requires a
modicum of treble cut, whilst the latter requires boost.
Now, move the microphone up or down and it is
apparent that the distances of the two drivers to the
Best results are found by ear.
20 21
microphone are no longer equal.
Figure 51
The result is that any phase correction in the crossover
is only valid directly on axis. In fact, any blending at all
of the two drivers suffers phase distortion off axis.
Not so with a Uni-Q combination; the two drivers
remain at the same point in space and have the same
distance to the microphone, whatever the angle of
measurement.
This is how the output of the speaker would look with
the phase correction switched OFF:
In addition, there is a dynamic bass stage that limits
the driver excursion whatever the setting. This
prevents excessive distortion and protects the driver
from possible damage. The application is practically
unnoticeable - however the effect is creating a
loudspeaker that sounds larger than it actually is.
Should the user prefer to be unbounded by the
limitations of a small bass driver, there is always the
option of adding one or two subwoofers.
Conventional speaker, mic. above axis.
Figure 52
Figure 53
Conventional speaker, mic. below axis.
Even measuring off axis in the horizontal plane results
in unequal path differences once the angle is large
enough that the radiation from the bass driver does
not travel directly to the microphone, but has to travel
rst to the front of the cone.
Figure 54
Conventional speaker, mic. horizontally off axis.
Figure 55
Uni-Q speaker on & off axis
Having established just how suitable a Uni-Q speaker
is for phase correction, let’s see how effective the
process is. Here is an input signal - a square wave of
1kHz:
Figure 56
Input signal - 1kHz square wave
Speaker output - Phase Correction OFF
Figure 57
And this is how it looks with the phase correction
switched ON:
Figure 58
Speaker output - Phase Correction ON
Bass Extension
There are three settings, which should be chosen
in conjunction with the room acoustics The “Less”
setting is similar to the natural bass extension of the
passive version, giving a cut-off frequency of 46Hz
(-6dB). The “Standard” and “Extra” settings give cutoff
frequencies of 43Hz and 40Hz respectively.
Normally, the “Extra” setting will be chosen, unless
there is a particularly bad room resonance around
30- 40Hz, where one of the higher cut-off settings will
compensate.
Adding a Subwoofer
Either one or two subwoofers may be added to
reinforce the bass of the audio system.
Each speaker (Primary or Secondary) has a Subwoofer
Out socket (RCA Phono). The signal sent to both
subwoofer outputs is a summed mono signal.
A crossover between the LS50 Wireless II and the
subwoofer is provided and the crossover frequency is
set within the app. There is no choice of lter shape - it
is always a 4th-order Linkwitz-Riley giving -6dB at the
cut-off frequency and a at in-phase total response.
If there is any low-pass lter supplied with the
subwoofer, it should be bypassed if possible or, if not,
the frequency should be set to the highest possible.
The choice of crossover frequency is something of a
compromise. The higher it is, the less the excursion
demands on the LS50 and the louder the whole
system can play. However, if a mono subwoofer output
is selected and the crossover frequency is too high,
some of the stereo information is lost. 80Hz is usually
considered the highest to avoid any perception of loss
of stereo information and the lower it is the better in
this respect.
In Basic setup mode, the cut-off frequencies of the
high-pass to the LS50 Wireless II and the low-pass
to the subwoofer(s) are always the same and this
is normally the course to follow. However, some
subwoofers do not have a at response and different
cut-off frequencies may give better results. Having
different frequencies requires the user to use Expert
mode for setup.
22 23
Should the subwoofer have an inverting amplier
(rare) or be sited well away from the LS50 Wireless
II such that the output is not in-phase with that of
the LS50 Wireless II at crossover, there is the option
to invert the signal to the subwoofer. In this case the
appropriate toggle should be set to ON. More usually
it is set to OFF. The correct setting will provide a fuller
sound.
If no subwoofer is used the high-pass lter to the LS50
Wireless II should be bypassed by switching the toggle
in the app to OFF.
The correct sub gain setting depends on several
factors such as the sensitivity of the subwoofer and
how many are used. The correct balance may be set by
ear.
rmware, to improve security, add functionality or
correct bugs.
Fortunately, the user needs to do nothing. If there is an
update available, it will be automatically downloaded
from our servers overnight, if the devices are either
fully on or in standby.
If preferred, users can manually download updates at a
time to suit them using the app. It is also not necessary
for the interspeaker cable to be connected.
The KEF Connect app will also receive updates, the
process of updating/setting automatic updates will be
dependant on the smart device.
Power Amplication
Summary
All too often, active speakers simply add power ampliers and active crossovers to a passive design, but there is
much more that active drive can offer.
This speaker takes an already excellent passive design and benets from all the acoustic technology packed into
it. The inherent benets of Uni-Q are improved upon even further through the inclusion of Phase Correction,
whilst the lack of a physical crossover eliminates what can be a signicant source of distortion.
LS50 Wireless II also represents a loudspeaker system equally driven by the concept of ‘Let You Be You.’
LS50 Wireless II does not aim to change the way someone experiences their music, but is designed to elevate the
individual’s unique situation through access to a plethora of music services and personalised EQ settings. Further
features and improvements, where developed, will be available through easy to perform rmware updates.
Syncing with Vision
Should the speakers form part of a home theatre
installation, where the programme content is
audiovisual, the thorny question of the sound syncing
with the picture (so-called lip-syncing) must be
addressed.
The all-pass characteristic of the bass/midrange to
tweeter crossover adds delay to the tweeter, so any
correction, if it is to be causal, must delay the whole
signal. Additionally, wireless transmission adds some
delay. It is important that all this inherent delay
does not lead to the perception of mis-syncing, even
if theoretically it is impossible to keep everything
absolutely together for all settings.
The Advanced Television Systems Committee (ATSC)
recommends that the sound to vision delay tolerance
is between -15ms to +45ms if lip-syncing is to be
unnoticeable. The maximum delay through the speaker
is comfortably less than 6ms and so lip-syncing should
not be a problem, whatever the settings chosen in the
app.
Firmware
From time to time, like any software-controlled device,
it will be necessary to update the LS50 Wireless II
As a true active loudspeaker, LS50 Wireless II features
four individual ampliers - one for each drive unit in
the system. These sit after the crossover, meaning that
each amplier receives only the frequencies destined
for the corresponding driver. This differs from a
passive or powered system, where the crossover
comes after the amplication.
A signicant benet from a design standpoint is that
different ampliers can be explored for each drive
unit. Through intensive computer modelling and
listening tests, KEF engineers were able to develop
amplication that is perfectly matched for the
frequencies - and drivers - at hand.
LS50 Wireless II features the same amplier classes as
the original LS50 Wireless - Class AB for the tweeter
and Class D for the midrange/bass driver. However, a
number of changes have been carried out, owing to
improved R&D and driver improvements.
The Class AB high frequency amplier sports a totally
new topology - not only providing a higher output,
but distortion and dynamic capabilities have been
further enhanced over the original. The Class D mid/
low frequency amplier has also seen a signicant
increase in power output.
24 25
Specication - LS50 Wireless II
Description
Drive Units
Frequency Range (-6dB) 47Hz - 45kHz (Less)
Frequency Response (±3dB)
Maximum SPL at 1m 108dB (full range, no subwoofer)
Harmonic distortion <0.4% 175Hz - 20kHz (90dB SPL, 1m)
Crossover frequency 1.9kHz
Power Input 100-240VAC 50/60Hz
Minimum Impedance 3.5Ω
Power Consumption 200W (operating)
Wireless HiFi Speakers
Uni-Q™ Driver Array
LF/MF: Nominal dia. 130mm (5.25 in) magnesium/aluminium alloy cone
HF: Nominal dia. 25mm (1 in) aluminium alloy dome
vented with Metamaterial Absorption Technology (MAT)
43Hz - 45kHz (Standard)
40Hz - 45kHz (Extra)
53Hz - 28kHz (Less)
48Hz - 28kHz (Standard)
45Hz - 28kHz (Extra)
< 2.0W (standby)
Power Ampliers LF/MF: 280W Class D
HF: 100W Class AB
Input Impedance 50kΩ
Dimensions H: 305mm (12.0 in)
W: 200mm (7.9 in)
D: 311mm (12.2 in)
Weight Primary: 10.1kg (22.2 lb)
Secondary: 10.0kg (22.0 lb)
Supported Formats (all inputs) MP3, M4A, AAC, FLAC, WAV, AIFF, ALAC, WMA, LPCM and Ogg
Vorbis
Supported Formats
(extra on network inputs only)
MQA
DSF: DSD64, DSD128, DSD256
DFF: DSD64
26 27
Loading...