What's new

 

HOME

 

Products

ORION - PLUTO
by Wood Artistry

ORION - EU

ORION - BR

ORION - PLUTO
by the sea

Do-It-Yourself

 

 

Concepts

Accurate Stereo

Loudspeakers

Crossovers

Room acoustics

Sound
reproduction

Room optimized
stereophonic
reproduction
 

Recording for stereo

Design of
loudspeakers

Issues in 
speaker design

 

 

Projects

ORION
challenge

ORION-3.1

PLUTO-2.1

PLUTO+
subwoofer

THOR
subwoofer

PHOENIX
dipole speaker

     Active filters

     Microphone

     FAQ

Reference
earphones

Surround
sound

Loudspeaker
evaluation

 

 

Resources

Website CD

Toneburst CD

Demo CD

Publications

Music

Links

Other designs

Site map

About me

 

 

Conclusions

 

 

HOME

 

--------------------
ORION
PLUTO
Users Group

 

 

--------------------
LINKWITZ
RENTALS

 

 

--------------------
Digital Photo
Process
es

 

 

What's new

 

 

Controlled directivity loudspeaker designs 

Loudspeakers in a domestic setting are used in rooms that are acoustically small in the bass region where the wavelengths are long (e.g. 7.8 m at 50 Hz) and which become increasingly larger as frequency increases and wavelength decreases (e.g. 3.4 cm at 10 kHz). At the low end of the spectrum there can be problems in the response due to room modes and at higher frequencies due to specular reflections. The room can participate in the sound reproduction process at many frequencies depending upon the absorptive and diffusive properties of surfaces and objects in the room and to the degree that these are illuminated by the sound waves emanating from the loudspeakers.

Loudspeakers tend to radiate uniformly into all directions at low frequencies where the radiating surfaces and cabinetry are small compared to the wavelength. They tend to beam the sound at high frequencies because the radiators are typically larger than 1/8th of a wavelength. Thus a loudspeaker, which is flat on-axis, radiates a different spectrum at angles off-axis, where the response is no longer frequency independent. Consequently reflections of the sound from different parts of the room that reach the listener will have different spectral content than the direct sound. The effect of reflections upon imaging from a stereo system has only been studied to a limited degree. It is often assumed that reflections can only have negative effects. This seems reasonable because reflections can only degrade the accuracy of a direct signal. But 2-channel stereo is about creating an illusion in the listener's mind. Neither the microphone signals, nor the ensuing ear drum signals at the listener, are accurate representations of a natural occurrence, of a natural hearing situation. The ear drum signals, though, contain cues from which the listener forms an auditory scene in his mind. If this scene is believable, then great enjoyment can be drawn from it.

The ear drum signals also contain cues about the room, though colored in the case of a loudspeaker that radiates a different spectrum in different directions. It has been my observation that a spatially more open and 3-dimensional auditory scene is created in one's mind when the loudspeakers radiate uniformly in all directions like dipoles or omnis do. The auditory scene has great clarity and distance but not the hard edges and closeness that a highly directional loudspeaker tends to produce

A dipole is the only radiator that is directional down to the lowest frequency. I have observed that it can produce bass output that sounds natural and effortless and unlike what I hear from most box speakers. A dipole source demands large volume displacements because of the phase dependent acoustic short circuit between front and rear radiation. The effect is wavelength dependent and reaches a minimum when the front-to-back distance D equals 1/2 wavelength, at which frequency the rear wave adds in phase to the front wave. Above this frequency the on-axis output decreases and becomes zero for D/l = 1. The radiation follows a cos(a) pattern for D/l < 0.1, but widens considerably above this. In practice this tendency is counteracted by the radiating surface becoming itself larger compared to the radiated wavelength and thus starting to beam. This helps when transitioning in a multi-way loudspeaker to a smaller dipole driver for higher frequencies. But it remains difficult to build a loudspeaker that has the same frequency response at off-axis angles as it has on-axis and where only the overall amplitude decreases with increasing angles. It can be done with small drivers, but then the output volume capability is compromised severely.

I do not know how consistent the radiation pattern has to be. Nor do I know that constant directivity over the whole frequency range is optimum. I do know that ORION and Pluto can deliver a very believable auditory scene and that other loudspeakers with different radiation patterns   

often create a less convincing illusion. The problem must be how a given loudspeaker system blends with the room. Ideally the room and the loudspeakers in it are not part of the auditory scene. That scene should open effortless and untiring in front of the listener, to be enjoyed best with closed eyes.

 

Other designs

On this web page I want to provide links to engineering information, to websites or forum discussions for the purpose of spreading learning and knowledge about controlled directivity loudspeaker designs. The material should be current and updated to reflect progress that has been made. I will need your help with that and invite you to give me links. The list would serve as a quick entry point to find out where and what the action is. It should therefore also point to relevant activities in languages other than English, since much can be learned from pictures and graphs.  

I expect that the list will point to different design approaches. It would be important to know the design objectives and to what degree they were met. To be added to the list please send me an email with the subject DESIGN LIST, with your URL, and with a 50 to 200 word statement of your specific Design Objectives. I will look at the website or forum to determine whether the loudspeaker design or some of its aspects are a contribution to the field of controlled directivity loudspeakers from my point of view and appropriate for the purpose of the list. 

--------------------------------------------------------------

 

Engineering information

Horbach-Keele linear-phase digital crossover filters for pair-wise symmetric multi-way loudspeakers - 6/9/10

   Here is a truly ground breaking, sensible and practical application of DSP to the design of crossover filters and the polar response of large multi-way active loudspeakers. Very exciting work! It avoids lobing of the vertical polar pattern by acoustically tight spacing of the driver pairs.

 

Constant beam-width transducer (Keele)  

Wide and uniform horizontal dispersion and controlled vertical dispersion up to high frequencies without lobes. The large number of small drivers ensures high output capability, especially in the tweeter range. The floor reflection is part of the design. The sound field is already uniform close to the loudspeaker and suitable for near-field listening. SPL falls off at a low rate with distance (3 dB/oct) and is nearly constant over a distance range. This is ideal for a home theater setup with rows of seats. (SL) 
Links

 

 

JBL Pro LSR6332

This loudspeaker exemplifies Floyd Toole's loudspeaker directivity requirements. They are the result of extensive listening tests where different box loudspeakers were ranked according to preference. The directivity index increases smoothly from 0 dB to 10 dB, without signs of the two crossovers in its frequency response. Moderately wide dispersion horizontally. (SL)

 

 

Genelec 8260A  

An active 3-way loudspeaker with a 10" woofer and coaxial 5" mid and 3/4" tweeter. Smooth transition from omni to +/-50 degree, -6 dB, horizontal dispersion. DSP controlled. (SL)

 

 

Directivity in loudspeaker systems (Geddes)

A 2-way loudspeaker with a 15" woofer and compression tweeter. Omni at low frequencies and rapid transition to a beam of +/-40 degree at -6 dB due to a 15" diameter waveguide. The narrow beam widens the sweet spot if the speakers are toed-in to cross in front of the listeners. The contour map would be even more illustrative of reality if it were drawn in polar coordinates with the frequency axis as radius and the angle covering 0 to 360 degrees. (SL)

 

The beneficial coupling of cardioid low frequency sources to the acoustics of small rooms (Ferekidis) 

ATC Loudspeaker Technology Ltd.

KEF Concept Blade

Perfect 8

 

DIY designs

2p Audio Pages - 4/1/10
Aristoteles and Platon are reference designs for me from which other projects will evolve.

Monte Kay - 3/19/10
- Envision a dipole with an acoustic black hole behind it, completely eliminating the rear wave, leaving only the front, frequency invariant lobe.  This best describes my directivity goal.  I utilize open baffle dipole and cardioids as tools to eliminate off axis radiation as a partial means of achieving this.  This design objective requires significant absorption behind the speaker to approximate the acoustic black hole.
- The “CBT” (Constant Beam width Transducer) as described by D. B. Keele has proven to be an effective means of achieving this goal.  My home theatre center speaker combines CBT technology with open baffle cardioids.  This focuses the rear wave at the center point of the CBT arc making it very easy to know where to put the absorption.  Along with the proper absorption, the open baffle CBT very effectively accomplishes my stated objective. 
- Keele and Horbach’s Linear Phase Symmetric Pair approach also accomplishes my objective but with other limitations.  The large surface area in the sum of numerous drivers in the CBT solves other problems not related to directivity making it a much higher performance design over the Symmetric Pairs.  As this is a discussion of directivity objectives, the other advantages are for another discussion.

keyser ( Martijn Mensink) - 3/17/10
- Flat frequency response, on-axis as well as off-axis by designing for a dipolar radiation pattern from the bass range up to the highest possible frequency. 
- Operating drivers largely below the first dipole peak to maintain constant directivity up to about 6.5 kHz. Small U-frame 12" woofers and no baffle at all for the 6" midrange and magnetostatic dipole tweeter.
- Closed box subwoofers will be added at a later date. The stand-alone dipole is currently equalized flat to a little below 40 Hz.
- Sufficient dynamic range. In practice it turns out that at a listening distance of about 3 meters and an average listening level a little over 80 dB(A) and an approximated source material crest-factor of 6 dB and broadband, spectrally dense content, there is no audible compression or distortion. At higher levels the sound becomes a bit congested, but I am not sure if this is caused by the room, the speaker or even my own hearing. This is sufficiently loud for me. I have not yet done any distortion measurements.
- Digital crossovers and equalization. Crossover frequencies are 300 Hz and 2000 Hz, both at 48 dB/oct.

John K - 3/14/10
When designing a speaker system for home use the objectives of any particular design will depend on the specifics of the application and acceptable trade offs. As such it is difficult to state categorically any specific set of design objectives for a CD speaker. With regard to constant directivity, my current interpretation would be that a CD speaker, intended for use in home environment, should have the ultimate goal of maintaining uniform polar response above the Schroeder frequency. As a rule of thumb, this translates to maintaining uniform polar response form about 100 Hz and above. Specific design objectives for my designs may be found at my website.

cuibono - 3/8/10
The primary objective is to develop an acoustically transparent loudspeaker that is involving, enjoyable and as life-like as possible.  This means addressing primarily linear distortion issues at all angles of radiation, and secondarily nonlinear distortion issues as they relate to maximum output levels.  This was obtained via the following goals:
1)  To be a full range dipole system.  In this case it is about 30Hz to 17kHz (F-6).
2)  To have as regular a dipole response as possible, defined as -1dB at 30°, -3dB at 45°, and -6dB at 60° relative to the driver's axis.  This design goal takes special attention above 1kHz, due to the midrange driver's basket structure, and the tweeters acoustically large size relative to the frequencies it is producing.  One compromise here is a limitation of output SPL from the midrange.  See post #35 in the Violet DSP thread for final polar measurements.
3)  To be as low cost as possible.  In this case, the total driver cost is about $400usd.
4)  To be have enough output SPL to play music at live levels, while keeping nonlinear distortion below an audible level.
(cuibono = Patrick Fleck)

MOB3W (my open baffle 3-way) - 3/6/10  
- Constant radiation up to at least 3kHz
- Symmetrical radiation across the entire spectrum
 
- Sufficient dynamic range down to at least 40Hz.
- Low distortion
 
I have made several baffles to test what happens with a midrange woofer on a baffle. I found that for true amplitude and phase symmetry, I had to sink the driver in the baffle. Each midrange has at the backside a construction directly behind the surround that has to be copied at the front. In addition, the slimmer the baffle, the more the radiation remains constant. This way the radiation of the AL130, the midrange in MOB3W, is made constant and symmetrical up to 2kHz. To obtain a radiation that is symmetrical and as much as possible constant for the tweeter, and close to figure-8 at least up to 3kHz, I designed the specific baffle of MOB3W. The AMT2340 tweeter 'sees' as little baffle as possible. It has ridiculously low distortion, not at all like the ESS AMT tweeters and the Eton ER4.

Gainphile - 3/4/10
Affordable lifelike reproduction of music. It is possible to build full-range dipole loudspeakers with lifelike reproduction capability at relatively low cost. The speakers are built under $500 and well under $1000 as complete systems including active 4-way analog crossovers and 8-channel amplifications. There is clear benefit on the accuracy of the drive signals by using active system. Measurable  transducer distortions and maximum SPL output are tradeoffs with such budget, yet the loudspeakers are uncolored and loud enough to provide listeners with enjoyable presentation.

StigErik - 3/4/10
True dipole operation over the entire frequency range. Operate all drivers below dipole peak to get better directivity. Decoupled drivers and/or baffles for reduced cabinet and/or baffle vibrations. Active XO and EQ. Choose drivers with good dynamic behavior and low distortion. Use multiple drivers to keep cone excursions far below Xmax. XO the midrange above 300 Hz so its less affected by the typical floor bounce suck-out. I'd like add that my listening room preferences and setup in the room differs somewhat from what is common. I have a LEDE room which kills most of the rear radiation from the dipoles (above 200 Hz). I also like to listen in the near-field - my current listening distance is just 1.8 meters (it should be rather obvious that I dont prefer to have early room reflections....) I also like to position the speakers at 45 degree angle instead of the usual 30 degrees.

 

Commercial designs

From some time ago - 8/4/10
A variety of significant speaker designs collected by Roger Russel of McIntosh Loudspeaker fame. 

TIMEDOMAIN  
Theory and technology behind small, single driver omnis. 

Stereolith  
A single box stereo loudspeaker with identical drivers on left and right sides and a single, mono tweeter on top. The L and R drivers on the sides are coupled via the internal air volume, which is a spring at low frequencies and becomes a transmission line for distances >
l/8 between the drivers. This causes dipole effects with peaks and dips at various angles. It is claimed that the brain handles such radiation favorably.

Grimm AUDIO  
A wide baffle 2-way loudspeaker with an IIR crossover, which imposes exact LR4 acoustic slopes crossing at 1550Hz. The phase is subsequently corrected using an idealized inverted all-pass filter, resulting in a maximally linear phase response without any pre-echo’s.

musicelectronic geithain gmbh  
RL 901K studio monitor with cardioid response in the bass region. No detail is revealed in a news interview how the two 30 Hz to 300 Hz flow-resistors have been constructed and actively equalized.

Georg Neumann, GmbH  
Directivity smoothly increasing from omni to a  +/-30 degree beam, forward horizontally

Steinway-Lyngdorf  
Dipole and boundary woofer

Philips  
Omni

Aether Audio  
Omni and low edge diffraction tweeter with low xo frequency 

Amphion Loudspeakers
Omni-cardioid-waveguide tweeter

Rountree acoustics  
Omni and forward radiating ribbon tweeter

jamo  
Dipole except for tweeter

Perfect 8  
Line dipole with ribbon tweeter

Gradient  
Cardioid with dipole woofer

BeoLab 5  
Omni at low frequencies and horizontal dispersion lens for highs

mbl 101
Omni 

GERMAN PHYSIKS
Omni 

WOLCOTT AUDIO  
Omni

Duevel  
Omni horn

RAAL requisite
Ring radiator

morrison audio
Omni 

Geddes Loudspeakers 
Omni at low frequencies, transitioning to narrow and constant directivity at highs

 

 

----  We came to find the forest. We got lost in the trees  ---- 
_________________________________________________________________
Last revised: 08/09/2010   -  © 1999-2010 LINKWITZ LAB, All Rights Reserved