KEF R3, R5, R7, R11, R2c User guide

KEF R&D

R Series 2018

2

Contents

Introduction 5
Philosophy 5
Objectives 5
The cabinet 6

Visual design 6

Minimising panel sound radiation 6

Directivity and diraction 8

Ports 9

Resonance inside the cabinet 10

Organ pipe resonances 10

Turbulence 10

Drive units (Drivers) 10

Bass drivers 10

Diaphragm 11

Magnet system 11

Spider (Damper) 12

Uni-Q array 12

Filling in the gaps 13

Midrange 13

Cone and surround 13

Magnet system 14

Tweeter 14

Diaphragm assembly 14

Tangerine waveguide 15

Rear loading 15

Crossovers 16

Model Range 18

R3 18

R5 19

R7 20

R11 21

R2c 22

R8a 23

Summary 24

Acknowledgements 24

References 24

3

R11 in Black Gloss nish

4

Introduction

Objectives

It is 7 years since the rst R Series products were introduced.

In that time, KEF’s Research and Development work has

continued and its best technology has improved. The fruits of

this work can be seen in the 2014 Reference Series and the

Blade 2. Behind the R series 2018 are a wealth of technical

updates and improvements – some trickle-down from the

Reference and Blade projects and some specic to the new

series. The result is a signicant step up in performance and

a purity of sound that’s clear from the rst listen.

Philosophy

“Of all art, music is the most indenable 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 artice.”

Raymond Cooke, KEF founder

Loudspeakers have a dicult job, not always fully appreciated.

They must take a 1-dimensional signal and turn it into a

3-dimensional experience.

Stereo – the simplest common conguration – uses just

two loudspeakers to create a complete sound stage for the

listener. Imagine throwing a handful of gravel into a pond and

trying to recreate the same ripple pattern by dropping just

two pebbles into the pond spaced by about 2 metres. It can’t

be done and we haven’t even considered height! Stereo is

an illusion that, fortunately, seems to work remarkably well.

Nevertheless, there are some clearly dened attributes that

separate good loudspeakers from mediocre ones:

The sound must come from the drivers and no other part of

the structure. The most common form of unwanted added

sound is radiation from the cabinet. The panels are relatively

large in area and don’t have to move much to colour the

true sound.

These words were penned shortly after KEF’s founding in

1961. Cooke was an avid music acionado and his mission

from the outset was to enable a wide audience to derive

the same pleasure from music as he did. He sought to

deliver to his customers the immersive experience of the live

performance through recordings.

Cooke realised that this could only be achieved by a scientic

understanding of sound and its reproduction, concentrating

his eorts on loudspeakers – the last link in the recording

chain and probably the most dicult in that they have to

work in an unknown 3-dimensional environment. Never

afraid to employ the most capable engineers and provide

them with the latest and most eective tools, Cooke

established a philosophy that still exists today.

KEF engineers continue to pursue this scientic endeavour,

today using tools and technology that were unavailable to

Cooke in those early days, never forgetting that all the science

in the world is useless unless it brings better performances

and greater pleasure to the user. The listening experience is

still the nal arbiter in deciding whether or not the science

is eective, but without it progress cannot be measured and

the way forward cannot be understood.

A loudspeaker should have controlled dispersion. If the

character of sound varies too much with angle, the

3-dimensional sound stage cannot be accurately formed

and an acceptable listening position is very restricted. But

omni-directionality is not desirable either. The acoustics of

the listening room become more intrusive and can swamp

any attempt to recreate the acoustic environment of the

original performance.

The sound quality should be the same at all levels. The

dynamic range of music (the quietest to the loudest sounds) is

high and, in order to reproduce it accurately, the loudspeaker

must not compress the sound nor to come to life only at

relatively high levels.

There should be no harmonic or intermodulation distortion

– the creation of sounds at frequencies that are not present

in the original.

There should be no time or temporal distortion. This can

come from resonances, where the sound hangs on at

discrete frequencies, or diraction, the re-radiation of sound

at discontinuities in the structure that is delayed relative to

the direct sound from the drivers.

5

Now, it’s all very well to say that there must be none of

anything, but we don’t live in a perfect world. The job of

the engineer is to minimise all these undesirables, ideally to

a level below which they are audible. However, sometimes

there are conicting issues and, in this case, the engineer

has to gauge the lesser of two evils. For example, low bass

at high volumes is extremely dicult for smaller cabinets to

achieve. However, smaller cabinets may be preferred for

their visual quality.

Visual design

All R Series cabinets are designed to look good. They have

clean lines, there is a choice of nishes – black gloss, white

gloss, walnut – and the frontal aspect is slim (more of that

later when we discuss temporal accuracy). The quality of

t and nish is high and, for those who prefer not to see

drivers, there is the option of magnetically attached grilles.

This brings us to the subject of visual design. Loudspeakers

must be considered a legitimate piece of furniture in the

living environment. They should, in short, look good.

The cabinet

We have seen that the objectives of the cabinet are:

• To look good.

• To emanate a minimum amount of sound.

• To maintain temporal accuracy.

• To help control directivity.

There is a fth attribute that the cabinet should also have and

it is stability. The most obvious criterion is that it should not

be easy to knock the loudspeaker over. Loudspeakers can

be heavy and could cause discomfort or injury should they

fall on anyone, especially a small child. There are furniture

standards that cover this and the device should return to an

upright position if it is leaned over up to 10°.

Also important to the audiophile is that the cabinet should

not be allowed to rock when the drivers operate. If this

happens, the bass quality is impaired. It may be described as

“softening” and the eect is most noticeable on percussive

bass, often referred to as “slam”.

All the oor-standers in the range are equipped with spike

feet that pierce through carpeting and prevent any rocking.

These spike feet are tted to outriggers that extend beyond

the boundaries of the cabinet and provide the necessary

static stability.

The loudspeakers are optimised for when the grilles are

not tted and they are ideally removed for serious listening,

even if they are tted when the loudspeakers are not in use.

However, the grilles are designed to have the minimum eect

on sound quality.

Minimising panel sound radiation

The panels of the cabinet may be excited by two mechanisms:

• Reaction forces from the drivers – as the driver

diaphragm moves back and forth there is a reaction

force on the chassis (basket) that is them transmitted to

the cabinet.

• Forces from pressure in the air contained inside

the cabinet.

If cabinets suer these problems, it is worth asking why we

need them at all. The answer lies in the drivers themselves.

They radiate the same amount of sound energy to the rear

as they do to the front. Unfortunately, the rear radiation

is out of phase with that from the front and, if the two

are allowed to mix, they cancel each other out. The

ability of sound waves to go around corners depends on

the wavelength relative to the size of the object. A good

analogy is to consider waves in the sea. They readily wrap

round small rocks but are reected from clis. So it is with

a loudspeaker driver. The lower the frequency the more the

sound will wrap around the driver. A driver suspended freely

in air will have no bass. We need a mechanism for blocking

this interaction and the cabinet serves to do this.

Stand-mount systems should always be mounted on a stand

that aords similar stability.

There are two common ways of reducing panel output:

• Adding bracing between the panels to increase stiness.

• Adding damping pads to the panels.

The idea behind method 1 is that the increased stiness

reduces panel motion. What it does in fact is to raise

the frequency of any resonances. A similar eect would

6

occur if the panel thickness were to be increased. This

might be of some benet if the loudspeaker system has

separate enclosures for dierent drivers, like in R Series.

The resonances in any one enclosure may be pushed out

of the working frequency range of the driver housed in

it. However, there is a drawback to this method. The bass

resonances may be pushed into the midrange, where the ear

is more sensitive.

The addition of damping pads to the panels does work and

was used on the R Series 2011. The pads are usually made

of some bituminous material that converts deformation into

heat and thus reduces the vibrational movement of the base

panel. Unfortunately, most commercially available pads are

designed to damp resonances in the thin metal panels used

on cars. Their ecacy depends on matching the mechanical

impedances of panel and damping pad and the match

between the pads and wood is not ideal.

KEF Engineers developed a technique, originally for the

LS50 loudspeaker system but applied to all R Series 2018

systems, that combines the two approaches and is more

eective than both. The technique, called damped bracing,

consists of strategically placed wooden braces that, instead

of being rigidly coupled to the panels are joined via a lossy

interface. The braces are placed where the panel movement

would otherwise be greatest and the lossy layer, like normal

damping pads, converts the energy into heat.

Figure 1

Driver in free air

Additional damped braces are situated at the back of each

driver, which reduce the level of reactionary vibration

transmitted from the driver to the cabinet.

The reduction in sound output from the cabinet varies

model by model, depending on the exact geometry, but

is typically of the order of 20dB, as shown in gure 3.

Cabinet sound radiation is highly distorted and, at the

resonance frequencies, causes time-smearing as the output

hangs on. The increase in clarity wrought by the damped

bracing is readily noticeable, is probably best in class for

single-enclosure multi-way loudspeaker systems and is

comparable to the best high-end systems.

Figure 2

R11 Cabinet construction

Figure 3

Panel vibration — without damping — with damped bracing

7

Directivity and diraction

Both these attributes are aected by the geometry of the

cabinet and so are lumped together here.

It is not often appreciated that the directivity of any

loudspeaker system is controlled as much by the geometry

of the cabinet as by the size of the drivers. In this, the

engineer is helped by the fact that listeners may occupy

positions that vary in the horizontal plane, but ear height

diers little. It follows that it is much more important to keep

the horizontal dispersion of the loudspeaker wide than is it

the vertical.

It was stated above that each of the R Series cabinets is slim.

Not only does this minimise the cabinet’s visual impact, but

it allows the wide horizontal directivity required to create

a stable sound stage for all listeners. The larger systems in

the range employ multiple small-diameter drivers rather that

a single larger diameter one precisely so the cabinet can

remain slim and maintain the desired horizontal directivity.

There is a small reduction in vertical directivity – but this is

not a problem at bass frequencies, where the wavelength

is long compared to the driver spacing. In the more critical

midrange, the Uni-Q array gives far better vertical directivity

than can be had from multiple vertically-displaced drivers.

Acoustic Impedances after Beranek - Note that it is the 20dB/decade

Figure 4

slope of the real part that aects the driver response in its working range

inaudible and the driver’s response smoother. However,

the radius of curvature must be of the same order as the

wavelength of the sound, the overall width of the cabinet

increases, making it more visually intrusive, and the cost of

manufacture increases considerably. So, whilst this technique

is often found on the most expensive systems, it is not

appropriate to the R Series.

Diraction – the bending of wave motion around corners –

can cause time smearing. Some of the sound energy generated

by the drivers travels along the front surface or bae of the

cabinet and, if it reaches an edge, re-radiates as shown by

gure 5. It’s rather like having small secondary drivers spread

along the edge of the cabinet. But this radiation is delayed

compared to the sound coming directly from the driver,

approximately by the time it takes to travel from the driver

to the edge. There is thus a degree of time smearing that

impairs the clarity of what the listener hears. Not only

that, but there is perturbation of the frequency response

because this secondary radiation goes in and out of phase

with the direct sound as the frequency and commensurate

wavelength vary.

The classic way of dealing with this eect is to round the

sharp corners of the cabinet. In simple terms, this can be

regarded as creating a multitude of edges, each with much

lower levels of secondary radiation. The total level is much

lower with the result that the time smearing becomes

With most manufacturers, the diraction eect is something

to be suered and not talked about. But the fact that we

are talking about it here indicates that KEF engineers have

successfully tackled the problem by a bit of lateral thinking

made possible by the fact that Uni-Q drivers are used

throughout.

It has been known for many years, and is illustrated by Leo

Beranek in his book “Acoustics”, published in 1954, that the

acoustic impedance of a driver (the interface between the

diaphragm motion and radiated sound) is smooth when

mounted both in an innite bae (as expected) and at the

end of a long tube the same diameter as the diaphragm.

This latter situation comes about because the dispersion of

the driver is such that it doesn’t acoustically “see” the edges

of the tube. The physics of this situation can be bent a little

and it can be shown that a driver mounted in a cabinet not

much wider than itself has a very low level of diraction. It’s

not perfect, but the level of diraction is much lower than it

would otherwise be.

8

Figure 5

Simulation of tweeter output

top – tweeter mounted directly in bae

bottom – tweeter in Uni-Q array with shadow are— direct — diracted

If we look at the drivers used in the R Series systems, we

can see that the bass drivers follow this requirement, but the

midrange and tweeter drivers do not. Normally, the tweeter

would be the most to suer because it is much smaller than

the width of the cabinet, which is determined by the diameter

of the bass drivers. However, in the Uni-Q driver (more of

which later), the tweeter res into an acoustic waveguide

formed by the midrange diaphragm and it is the diameter of

this relative to the width of the cabinet that determines the

level of diraction.

Ports

Ports are used to augment bass response. They have the

dual advantage that they enable a lower cut-o 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.

• 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 chung).

KEF engineers have tackled each of these potential problems.

There is more work to be done, because the diameter of the

midrange (for other reasons) is smaller than that of the bass

drivers, and it is here that the lateral thinking comes into play.

Around the Uni-Q driver array is what is normally called a

trim ring. Not simply a decorative piece, the R Series trim

ring is specially shaped and engineered. It increases the

eective diameter of the waveguide such that the driver

array “sees” the edges of the cabinet much less than if it

were mounted normally. Hence the name “Shadow Flare”.

The level of secondary radiation is decreased, resulting in

less time-smearing, greater clarity and a smoother response.

This technique was originally developed for the Reference

Series products and illustrates our trickle-down philosophy.

Simulation of bass driver + port — and port only —when the port is

Figure 6

placed close to the node of internal resonances.

Figure 7

Simulation of bass driver + port — and port only —when the port is

placed close to the antinode of internal resonances.

9

Pressure of the rst “organ pipe” resonance in a port Top –

Figure 8

rigid walls and no end areBottom – exible walls and end ares

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 LS50

loudspeaker, this technique dramatically reduces the pressure

variations inherent in these higher frequency resonances and

renders them less audible. Figure 8 illustrates this technique

using Finite Element Analysis (FEA), where the colour scale

from green through to red indicates the level of air pressure.

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. The

progressive expansion of the air aorded by the aring

reduces turbulence and thus audible chung. 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.

Resonances inside the cabinet

The port(s) are 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.

• There is more freedom on the rear panel compared

to the front in positioning the port so it is placed close

to antinodes (nulls) of the resonances, so less energy is

transmitted through the port.

To illustrate this last point, gures 6 & 7 show the dierence

in output when the port is placed at a node and an antinode

of an internal resonance. For the rst of these measurements

there is no wadding inside the cabinet, so the eect 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.

Drive units (Drivers)

All but one of the systems are 3-way – they have one or

more bass drivers, a midrange driver and a tweeter. The

midrange driver and tweeter are combined into a single

array known as Uni-Q. The only 2-way in the series utilises a

bass-midrange driver instead of a dedicated midrange driver

in its Uni-Q array.

Bass drivers

Whether the system contains one, two or four bass drivers,

the design remains the same, save for the impedance of

each individual driver. The number of bass drivers simply

determines the maximum loudness of the system for a given

level of distortion. Note that, when there is an even number

of bass drivers, these are symmetrically arranged above and

below the Uni-Q array to preserve a virtual point source

over the full range.

10

Diaphragm

This is a 2-part structure, comprising a shallow concave

aluminium front skin connected to the driving voice coil by a

more conventional paper cone (see gure 9).

Cross-section of bass driver moving parts

Figure 9

The aluminium skin has a shallow prole to minimise

perturbation of the response of adjacent drivers as a result

of diraction.

The cone itself is vented to allow free movement of air as

the diaphragm assembly moves back and forth. The static

centre pole of the magnet structure causes the volume of

air between the paper cone and the front aluminium skin

to vary with diaphragm position and this must not become

trapped, otherwise the air will be pumped through the

narrow apertures of the magnet system and cause noise.

— Response of non-optimised bass driver — Response of optimised bass driver Panel

Figure 10

is not excited.

Magnet system

With the new series, the magnet system has been totally

redesigned. One of the key objectives is to keep the magnetic

ux density seen by the voice coil as constant as possible

with coil position.

The front aluminium skin provides the stiness necessary to

provide pistonic radiation and therefore temporal accuracy

within the working range of the driver. However, like all sti

materials, the stiness is not innite and at some frequency

diaphragm breakup sets in and the Q (sharpness) of the

resonances is high – the response peaks at resonance can

be as much as 20dB above the pistonic response. Such

resonances are not readily suppressed by the relatively

gradual attenuation of the crossover network and, if left

unchecked, will cause colouration.

To reduce the intrusion of these higher-frequency resonances,

the paper cone is joined to the aluminium skin at a nodal

ring, not at the periphery. The nodal ring is an antinode of

the lowest resonance and, when driven there, the resonance

Figure 11

Finite Element Analysis (FEA) of

Top – 2011 R Series bass driver magnet system

Bottom – 2018 R Series bass driver magnet system

11

To this end, the old magnet system featured an extended

centre pole that balanced the solid material that connected

to the back plate. However, this design is still somewhat

asymmetric, with the result that the force on the voice coil

was also asymmetric, introducing more distortion than is

necessary. Further, the asymmetric nature of the metalwork

caused the coil’s inductance to vary with position, altering the

termination, and therefore the equalisation, of the crossover.

The new design is more complex and features a so-called

T-shape pole (see gure 11).

This is more symmetric than the older design and, with the

bulk of the centre pole being further from the voice coil,

lowers both the coil’s absolute inductance and its variation

with position. The result is a useful reduction in even-order

harmonic distortion.

Note, too, the undercut in the back plate that allows greater

coil movement before the coil hits the back plate.

Another feature of the new design can be deduced by

looking at the magnetic permeability in the metalwork.

The darker the colour, the lower the permeability and, where

it is almost black, the metalwork is saturated. This has two

advantages for the performance:

• The saturation limits the ux density in the coil gap and

so this remains constant regardless of the inevitable

production tolerances in magnet strength.

• Saturation further reduces the absolute level of inductance

of the voice coil and, with it, further educes the variation

in ux density that would normally occur as the coil

moves back and forth in the gap.

Spider (Damper)

The spider, or rear suspension, contributes to the linearity

of coil movement in addition to the magnet system. In this

case, as we’re dealing with symmetry, it’s the odd-order

harmonics that are mostly aected. The new spider design

allows greater linear movement and a commensurate

reduction in distortion, clearly illustrated in gure 12.

Suspension compliance as a function of diaphragm displacement

Figure 12

— Old spider — New spider

Uni-Q array

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 suered by a true single driver covering such a wide

frequency range.

Over the intervening years, the Uni-Q concept has been

progressively rened. 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. This is the

12th generation of the array and brings an extra degree

of renement to the performance, but it is worthwhile to

describe everything that goes into this remarkable array, not

just the latest renement.

12

One of the requirements of building this driver combination

is the need to maintain tight tolerances. Some of the gaps

between parts are necessarily small and without robotic

assembly methods it would be impossible to achieve

consistency of performance. There is a place for hand

building, but this is not it.

Filling in the gaps

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 o – simulating a perfectly smooth waveguide.

Obviously, this annular gap is necessary to allow the midrange

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. Adding damping to this

newly-created cavity was found to be eective in taming

the resonances in the annular gap and the removal of

the response glitches was immediately apparent as an

improvement in detail clarity.

Figure 15

Simulation of tweeter response

— In perfect waveguide — With open gap

Uni-Q array with no gap between HF and MF magnets

Figure 13

Figure 16

Actual tweeter response

— Original — With cavity damping

Midrange

Cone and surround

The cone is formed from a magnesium/aluminium alloy. Like

the front skin of the bass driver, it serves to provide the

necessary stiness to give pure pistonic motion over the

driver’s working range. The stiness is increased by the radial

embossing in the cone prole, but nevertheless it is prone

to high-Q breakup in the frequency range covered by the

tweeter.

Uni-Q array with opened cavity between HF and MF magnets and damping added

Figure 14

Figure 17

Uni-Q drive exploded view

13

The KEF Blade uses nodal drive (as described on page 11,

where the technique is used on the bass drivers) to tame

these resonances. But that technique, when applied to a

Uni-Q array, cannot use an intermediate cone and instead

requires an expensive large diameter voice coil which was

not appropriate to the R Series.

Instead, there is a lossy interface where the voice coil joins

the cone. This interface serves to decouple the drive at high

frequencies. 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.

Response of midrange driver measured without crossover in 2π bae

Figure 19

— without interface — with interface

Figure 19 illustrates the improvement in the unit’s response

as a result of using the lossy interface.

The surround is particularly problematic. It has to allow

the cone to move freely, but it is also a critical part of the

tweeter waveguide. Surrounds also tend to break into

complex vibrational behaviour at higher frequencies, often

moving in the opposite direction to the cone. The so-called

“surround dip” in a driver’s response is quite common.

The new R Series driver features an improved surround

geometry, achieved through extensive computer simulation,

optimisation and practical experimentation. It allows greater

movement of the cone and gives a smoother response for

both midrange and tweeter.

Magnet system

The design of the magnet system follows that of the bass

drivers. alongside the addition of aluminium de-modulation

rings. The reduction in harmonic distortion as a result of

the new magnet system and improved surround geometry

is shown in gure 20.

lossy interface between voice coil and midrange cone

Figure 18

Midrange driver distortion (1m, 5.6V RMS)

Figure 20

— 2011 R Series — 2018 R Series

Tweeter

Diaphragm assembly

The dome diaphragm itself is aluminium. More exotic

materials – diamond and beryllium for example – are

sometimes used in an eort to increase stiness 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.

14

The optimum dome shape for waveguide loading is a

spherical cross-section, but the optimal shape for stiness

is an elliptical cross-section. Both were combined into the

patented KEF Stiened Dome. An elliptical dome attaches

to the voice coil former and the centre is removed. This is

capped by a spherical dome and a very sti triangular section

is formed where the two parts join.

Figure 21

3D CAD sectional views of the tweeter dome and extended former

that meet to form a triangular stiening member at the dome edge.

Tangerine waveguide

Figure 23

Simulated response of tweeter

— without tangerine waveguide — with tangerine waveguide

Rear loading

Like any drive unit, the tweeter radiates as much energy

backwards as it does forwards. Not only does this energy

have to be absorbed, but there must be sucient volume

behind the dome to prevent a build-up of excessive pressure.

This pressure can impair the suspension of the dome and

cause displacement of the ferrouid that sits in the magnet

gap to cool the voice coil.

Figure 22 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 non-ideal

situation and taking a leaf out of the design of pressure 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 23.

To that end, the dome vents through the centre pole of

the magnet system to a cavity lled with absorbent wadding.

The cavity is tapered to that, when the wadding is inserted,

it becomes more dense towards the rear of the cavity,

gradually increasing absorption.

All the rearward energy is absorbed and cannot re-radiate

through the dome after being reected at the back of

the cavity.

Figure 24

Front and rear view cross sections of tweeter showing rear loading

15

Crossovers

All 3-way crossovers, which covers all bar the 2-way Atmos®

enabled R8a, follow the same general format shown in gure

25.

Each system in the range has been carefully voiced using a

wide variety of musical genres to ensure that no articial

enhancement is introduced.

All cored inductors in the new series crossovers use a

laminated core material that substantially lowers distortion

compared to the outgoing series, as shown in gure 27.

Mention should be made regarding the two parallel

capacitors C3 and C4. The total value is large and low-loss

types would be prohibitively expensive, so the bulk of the

value is made up of an electrolytic type and around 10% of

the total value is a polypropylene type, wired in parallel - the

so-called bypass conguration.

All crossovers have a reduced component count. This has

been made possible due to the decoupler in the Uni-Q

midrange as well as the improved tweeter, both of which

have inherently smoother responses and require less

correction from the crossover. The series resistors in the

MF and HF sections are merely to change the output level of

these drivers to match that of the bass drivers.

The HF section features a tuned circuit (L1/C2) in parallel

with the driver. This tuned to the fundamental resonance

frequency of the driver and allows the series capacitor to

have the correct attenuation. Without it, there would be

excessive movement of the diaphragm and an unwanted

hump in the response.

There are two tuned circuits in the LF section. One (L6/

C6/R4) is tuned to the upper peak of the reex impedance,

and prevents the system impedance falling too low should

the 2nd-order low-pass formed by L4 and C5 be allowed

to interact with it. The inductor L5 is small and tunes with

C5 to attenuate a residual high Q resonance in the driver

response to prevent it colouring the overall sound.

This conguration delivers almost the same low loss and

quality that would be achieved by having the full value solely

supplied by one or more polypropylene capacitors wired in

parallel, but at a much lower cost.

Figure 26

Inductor distortion at 100W

— 2011 R Series — 2018 R Series

General schematic of R Series 3-way crossovers

Figure 25

16

17

Model Range

R3

Description

This is the smallest of the ‘regular’ loudspeakers in the range,

being a compact 3-way stand-mount model. It is suited to

the smaller room or closer listening, where less power is

required to reach realistic sound levels.

Technical Specications

Design Three-way bass reex

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF: 125mm (5in.)

aluminium cone

Bass Driver:

LF: 165mm (6.5in.)

hybrid aluminium

Crossover frequency 400Hz, 2.9kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 58Hz - 28kHz

Harmonic distorition <0.3% 130Hz - 20kHz

Maximum output 110dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.3.2Ω)

Sensitivity (2.83V/1m) 87dB

Weight* 13.5 kg (29.8 lbs.)

Dimension (H x W x D)

with terminal*

Finishes Black Gloss / White Gloss

38Hz - 50kHz

30Hz

15 - 180W

422.2 x 199.6 x 335.5 mm

(16.6 x 7.9 x 13.2 in.)

/ Walnut

* Measurement per unit

18

R5

Description

This is the smallest of the three oor-standers in the range.

It’s twin 130mm diameter bass drivers are positioned each

side of the Uni-Q array in d’Appolito conguration, to

preserve an eective point source delivery throughout the

whole frequency range.

Technical Specications

Design Three-way bass reex

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF: 125mm (5in.) aluminium

cone

Bass Driver:

LF: 2 x 130mm (5.25in.)

hybrid aluminium

Crossover frequency 400Hz, 2.9kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 52Hz - 28kHz

Harmonic distorition <0.3% 120Hz - 20kHz

Maximum output 110dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.3.2Ω)

Sensitivity (2.83V/1m) 87dB

Weight* 27.3 kg (60.2 lbs.)

Dimension (H x W x D)

with terminal*

Dimension (H x W x D)

with terminal with Plinth*

Finishes Black Gloss / White Gloss /

38Hz - 50kHz

29Hz

15 - 200W

1025 x 175 x 343.5 mm

(40.4 x 6.9 x 13.5 in.)

1071.4 x 271.6 x 343.5 mm

(42.2 x 10.7 x 13.5 in.)

Walnut

* Measurement per unit

19

R7

Description

Like the R5, this is a oor-standing loudspeaker featuring twin

bass drivers positioned above and below the Uni-Q array.

However, the larger diameter of the bass drivers enables a

higher sound level to be delivered and the larger cabinet

aords a greater bass extension.

Technical Specications

Design Three-way bass reex

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF: 125mm (5in.) aluminium

cone

Bass Driver:

LF: 2 x 165mm (6.5in.)

hybrid aluminium

Crossover frequency 400Hz, 2.9kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 48Hz - 28kHz

Harmonic distorition <0.3% 120Hz - 20kHz

Maximum output 111dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.3.2Ω)

Sensitivity (2.83V/1m) 88dB

Weight* 31.4 kg (69.2 lbs.)

Dimension (H x W x D)

with terminal*

Dimension (H x W x D)

with terminal with Plinth*

Finishes Black Gloss / White Gloss /

33Hz - 50kHz

27Hz

15 - 250W

1062 x 200 x 383.5 mm

(41.8 x 7.9 x 15.1 in.)

1108.9 x 310.6 x 383.5 mm

(43.7 x 12.2 x 15.1 in.)

Walnut

* Measurement per unit

20

R11

Description

This, the largest loudspeaker in the range, features four

165mm (6½ inch) diameter bass drivers, enabling a enough

sound level output to drive large rooms without distress.

The large cabinet aords the most extended bass response.

Technical Specications

Design Three-way bass reex

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF: 125mm (5in.) aluminium

cone

Bass Driver:

LF: 4 x 165mm (6.5in.)

hybrid aluminium

Crossover frequency 400Hz, 2.9kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 46Hz - 28kHz

Harmonic distorition <0.3% 120Hz - 20kHz

Maximum output 113dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.3.2Ω)

Sensitivity (2.83V/1m) 90dB

Weight* 37.7 kg (83.1 lbs.)

Dimension (H x W x D)

with terminal*

Dimension (H x W x D)

with terminal with Plinth*

Finishes Black Gloss / White Gloss /

30Hz - 50kHz

26Hz

15 - 300W

1249 x 200 x 383.5 mm

(49.2 x 7.9 x 15.1 in.)

1295.5 x 310.6 x 383.5 mm

(51 x 12.2 x 15.1 in.)

Walnut

* Measurement per unit

21

R2c

Description

This is the rst of two special application loudspeakers. It’s

a dedicated centre-channel model and, like the R5 features

two 130mm (5¼ inch) diameter bass drivers each side of

a centrally placed Uni-Q array. Normally, the horizontally

disposed drivers of centre-channel loudspeakers suer a

restricted horizontal dispersion in the crossover region

between midrange and tweeter, or throughout the midrange

band if two drivers are used to cover that frequency range.

However, the point-source Uni-Q array avoids this and oers

better horizontal dispersion than any other horizontally

disposed conguration.

Technical Specications

Design Three-way closed box

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF: 125mm (5in.)

aluminium cone

Bass Driver:

LF: 2 x 130mm (5.25in.)

hybrid aluminium

Crossover frequency 400Hz, 2.9kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 74Hz - 28kHz

Harmonic distorition <0.3% 120Hz - 20kHz

Maximum output 110dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.3.2Ω)

Sensitivity (2.83V/1m) 87dB

Weight* 16.9 kg (37.3 lbs.)

Dimension (H x W x D)

with terminal*

Finishes Black Gloss / White Gloss

64Hz - 50kHz

49Hz

15 - 200W

175 x 550 x 308.5 mm

(6.9 x 21.7 x 12.1 in.)

/ Walnut

* Measurement per unit

22

R8a

Description

This compact model is the only 2-way loudspeaker in the

range. It features a single Uni-Q array, this time having a bass/

midrange driver surrounding the centrally placed tweeter

unit. It’s angled front bae makes it the perfect solution for

either a wall-mounted surround loudspeaker or a speaker-

top-mounted Dolby Atmos® module.

Technical Specications

Design Two-way closed Box

Drive units Uni-Q Driver Array:

HF: 25mm (1in.) vented

aluminium dome

MF/LF: 130mm (5.25in.)

aluminium cone

Crossover frequency 2 kHz

Frequency range (-6dB)

Typical in-room bass

response (-6dB)

Frequency response (±3dB) 105Hz - 18.5kHz

Harmonic distorition <0.3% 200Hz - 20kHz

Maximum output 106dB

Amplier power

(recommended)

Nominal impedance 8Ω (min.4.2Ω)

Sensitivity (2.83V/1m) 86dB

Weight* 4.45 kg (9.8 lbs.)

Dimension (H x W x D)

with terminal*

Finishes Black Gloss / White Gloss

96Hz - 19.5kHz

-

25 - 150W

173.8 x 174.6 x 259 mm

(6.8 x 6.9 x 10.2 in.)

* Measurement per unit

23

Summary

References

The R Series sits immediately below the Reference Series

and borrows much of that range’s technological attributes.

It follows our belief that a great sounding loudspeaker is the

result of solid, careful and eective engineering. This involves

utilising every computer simulation and measuring technique

available, backed up by practical experimentation and

verication.

But loudspeakers, like any other component of the

reproduction chain, are there to communicate more than

mere measurements and specications. As Raymond

Cooke realised from the start, emotion and involvement

are also part of what must be delivered to the listener. For

that reason, engineers and non-engineers alike audition

prototypes at every stage of development to make sure

that the engineering is relevant and improves the overall

performance.

More than that, it is the same team that guides every KEF

product through its development, with the result that there

is a consistency of what constitutes quality throughout

the company’s portfolio. The products deliver outstanding

performance whatever the musical genre and regardless

of whether the programme is 2-channel or multi-channel

based.

The R Series incorporates many technologies that were

pioneered by KEF - the Uni-Q point source array is probably

the best-known and certainly one of the most eective

of these. But this paper illustrates how every aspect of

loudspeaker performance has been assessed and taken to the

limit, within the imposed budgetary constraints, sometimes

involving a new and dierent way of addressing the problem.

O J Pedersen, S Bech and K Rasmussen, Final Report on

Activities of Eureka Project ARCHIMEDES. Univ., 1995.

L R Fincham, A Jones and R H Small, “The Inuence of

Room Acoustics on Reproduced Sound, Part 2: Design of

Wideband Coincident-Source Loudspeakers”, in The 87th

Convention of the AES, 1989.

“The Reference (white paper)”, KEF R&D, May-2014.

K Kessler, D A Watson and M Cooke, “KEF 50 Years of

Innovation in Sound” – A Limited Edition Book, GP Acoustics

International Limited, Maidstone, UK, 2011.

M Dodd and J Oclee-Brown, “A New Methodology for the

Acoustic Design of Compression Driver Phase Plugs with

Radial Channels”, presented at the Audio Engineering Society

Convention 125, 2008.

M Dodd, “Optimum Diaphragm and Waveguide Geometry

for Coincident source Drive Units”, presented at the Audio

Engineering Society Convention 121, 2006.

M Dodd and J Oclee-Brown, “Design of a Coincident Source

Driver Array with a Radial Channel Phase-Plug and Novel

Rigid Body Diaphragms”, presented at the Audio Engineering

Society Convention 127, 2009.

M Dodd, W Klippel, and J Oclee-Brown, “Voice Coil

Impedance as a Function of Frequency and Displacement”,

presented at the Audio Engineering Society Convention 117,

2004.

Some of the techniques described here are dealt with in

greater detail in the KEF Reference Series white paper and

the interested reader is urged to peruse that in addition to

the other publications featured in the list of references on

the next page.

Acknowledgements

The KEF logo and Uni-Q is are registered trademarks of GP

Acoustics (UK) Ltd. Dolby and Atmos are trademarks of

Dolby Laboratories.

L L Beranek, “Acoustics”, McGraw-Hill, 1954 revised 1986

G Perkins, “Modelling loudspeaker cabinet diraction”,

presented at the IOA Reproduced Sound conference, 2014.

M Dodd, “The Development of the KEF LS50, a Compact

Two-way Loudspeaker System”, presented at the Audio

Engineering Society Conference: 51st International

Conference: Loudspeakers and Headphones, 2013.

24