And the next speaker is...!

As noted, in the early period of loudspeaker development, the 'reed', or moving iron speaker was still believed to be a fruitful solution - but units typified by the early 'Amplion Lion', speaker were essentially telephone receivers with large paper cones driven by a very small (low mass) steel armature near a coil-wound permanent magnet, and suffered badly from mid-frequency resonance. But like the rest, the 'Amplion Lion' system was very inefficient and had a poor mid-to-high frequency response.

 

 In 1919 C W Hewlett brought out a new plane diaphragm speaker. The diaphragm was suspended between two pancake coils where - operating in a permanently energised fashion - the coils created a push-pull action on the diaphragm;  imitated in modern configurations as two moving coil units face to face and operating in anti-phase.

 

Dynamic Loudspeakers

 

It was all rather empirical. For a start, no one had any real idea of the factors that made an ideal dynamic loudspeaker. As we have noted, most up to the early 1920's were horribly deficient - they were short on frequency range, had horrible resonances, were unable to handle transient signals and thus had a poor dynamic response.

 

They were insensitive and inefficient - efficiency and sensitivity were correlated. Low field magnet strength meant that higher voice (motor) coil currents were required for a higher acoustic output. The same output could be produced with lower currents if the strength of the field magnet could be increased. Nevertheless, most examples were tolerable when it came to reproducing the relatively poor signals they had to contend with.

Related further reading...

Further Reading:

Part 2

Barrie Blake-Coleman

To counter inherent defects, and make the most of better quality audio signals, the transducer had to reproduce the sound without distortion or limitation of the audio frequency range. Furthermore, they had to do this at a reasonable power level and with a good dynamic range.

 

A New Understanding

 

The foremost, and most significant, attack on the problem was due to two Americans - Chester W Rice and Edward. W Kellogg working for General Electric Research Laboratories. Their seminal work on elucidating the behaviour and ideal properties of the dynamic loudspeaker was published in 1925 (Notes on the Development of a New Type of Hornless Loud Speaker. Trans AIEE, Vol.44, pp. 461-475 April 1925). They disclosed a number of important aspects of loudspeaker design. First, that a moving coil and its suspension, driving a diaphragm or cone, was a spring mass system with a 'natural frequency' of simple harmonic motion (mechanical resonance).

 

The electrical energy delivered to the voice coil was expended in accelerating the mass of the system (inertia control). In electrodynamic terms, this meant that the moving mass (diaphragm/coil and suspension) was equivalent to electrical inductance, and the spring component (the combined coil suspension, diaphragm 'compliance' and coil self-inductance) was equivalent to electrical capacitance. In short, the system was the mechanical equivalent of an electrical resonant circuit having a certain fundamental frequency. All other factors being equal, the system was mass controlled.

 

It followed that if the mass spring ratio was right, it was possible to design a system where the fundamental resonance was below the lowest frequency of interest. Furthermore, it was also possible to create a resonance which accentuated a particular frequency of interest. Important too was the acoustic radiation properties of the loudspeaker.

 

Rice and Kellogg established (1) that the acoustic power output of a loudspeaker was the product of the sound radiation resistance (Qs) and the square of the displacement velocity (V2) of the cone/diaphragm (P=QsV2). Likewise, (2) the radiation efficiency of a small diaphragm introduces a mechanical resistance (Rm) which varies as the square of the frequency (Rm proportional to f squared).

Larger Cone Areas

 

It was apparent that lower frequencies were more efficiently radiated if the cone area was large, even though the displacement was low. Verso, the cone displacement could be large with a corresponding reduction in surface area. There was, nevertheless, a difference in approach with a distributed driving force - that is, where the motional force was not at the vertex of a cone alone but over the whole diaphragm area. Such designs were better low frequency radiators -the induction unit of C W Hewlett was cited - so too the Gaumont designed small cone loudspeaker made by Societe des Establissements Gaumont in France. In a coil driven, flexible cone loudspeaker, the radiation frequency was determined by the cone elasticity; that is, higher frequencies emanated through flexural waves radiating from the centre section, lower frequencies radiated from the outer.

 

Ultimate Wave Action

 

This wave action occurred from the centre radius of the cone to the outer circumference. At very low frequencies the cone behaved more as a piston, but if the wave attenuation (energy loss) across the cone (from centre to outer) was too great (typically in a very thin, highly inelastic cone) the higher frequencies, above the piston frequency and the outer cone radiation frequency, would be accentuated. But even this was a problem.

Get the cone mass and elasticity wrong and the cone was apt to 'break up', producing subsidiary resonances at certain mid or high frequencies. Furthermore there was phase distortion, caused by the fact that the sound wave took time to travel from the cone centre to the periphery. Thus the cone electively radiated different frequencies at different times - but over a small area this phase loss was negligible.

 

Experimentally, and design wise, the 45°, 0.007 - 0.01 inch composite paper cone, suspended on a linear rubber surround, offered the best possible compromise between low frequency piston action, with mid to high frequency flexural wave action. Thus producing a more uniform response.

 

Where the resonance rule is observed, the complementary factors given in the first and second rule allow a (near) uniform response up to the frequency at which the assumptions begin to fail (2). Such revelations were the main contributions of Rice and Kellogg, and remain the prescripts on which all loudspeaker technology is based.

More to Know

 

Later work by other investigators (eg Olson and Klipsch etc) took the Rice and Kellogg work further and showed that the speaker mechanical/electrical properties need not always be built in, and that the use of appropriate baffles and enclosures introduced predictable mechanical behaviour which allowed speakers to perform better.

 

Radiation of a circular diaphragm in a baffle had been first elucidated by J W Strutt (Lord Rayleigh) in 1878 and C W Rice subsequently showed that a small free standing speaker cancelled the lower frequencies because the front to back pathway was too short, so with a path length approaching the distance of one wavelength pressure nulling (destructive interference) of the sound wave occurred. Increasing the path/distance critically meant that the lower frequencies were radiated properly, or rear and front radiation was reinforced through constructive interference, but this also meant large area baffles (mounting surrounds). The only alternative being very large cones (two feet plus!) with a proportionally higher low frequency radiation.

 

Stiff inner and outer cone suspensions (frequently at this time made from a paxolin 'spider' around the coil/cone joint) made for a much higher speaker fundamental resonance, and this could be usefully employed to accentuate bass harmonics rather than bass fundamentals - as such, these units were not in need of sophisticated enclosures and worked well in simple boxes typical of those used for AM radio sets.

 

Fundamental Resonance

 

High compliance speakers were designed for transmission path or bass reflex enclosures, where the greater part of the system stiffness was controlled by the pressure variations determined through the use of enclosure ports or acoustic labrinths . Such systems had very low fundamental resonances and were able to handle high powers with reduced distortion. As early as 1927 Wente and Thurus had demonstrated a dynamic speaker some 33% efficient. By 1937, the Williamson hi-fi amplifier (8 Class A Triodes) and state of the art speakers were reproducing shellac disc recordings to their optimum.

 

But the moving coil speaker was always in danger of being superseded. It was prone to amplitude distortion -anything displacing back and forth in sympathy with a varying signal was in danger of overshooting if the system inertia (mass displacement) was not perfectly controlled. Also, it was necessary to make the field gap as long as possible. It was evident that as the coil moved, its movement would become non-linear, and distortion would set in, if the magnetic field changed in strength over the path length. This would happen anyway at the maximum point of travel - but good speakers would have as large a displacement necessary to match power input requirements to a perfectly uniform annular field. Initially mains energised magnets solved the low field strength of permanent magnets - but with the risk of the field current being modulated with mains borne interference or AC hum. Comparatively poor materials discouraged better designs - extended travel in small cones was hard to achieve. As noted, an easier approach was a large surface area with a much lower displacement, but big speakers were impracticable and were poor at high frequencies.

It was all a compromise - which attempted the theoretical ideal for a wide frequency unit, that of an infinitely stiff, infinitely elastic, massless diaphragm acting as the perfect acoustic black body (uniform acoustic radiator at all frequencies) with a 100% electro-acoustic conversion efficiency and zero distortion. A virtually impossible specification with real materials.

Alternatives

 

So the dynamic moving coil system, whatever its advantages, was viewed as fundamentally flawed. It was a trade-off. If it was ideal at low frequencies it suffered in the mid to high range -though Rice and Kellogg had shown that this could be mitigated if the voice coil resistance was higher than the reactance, or if a copper ring was introduced to reduce the impedance at high frequencies. Still, if a speaker was optimised for high frequencies it tended to be poor at radiating the low. Piezoelectric crystal transducers were suitable for high frequencies as too the ribbon speaker (the reverse of the ribbon microphone - first developed by Siemens and Halske in 1922). Much criticised, for poor low frequency response, the ribbon speaker was originally designed as a 'full range unit' (Wireless Review and Radio World July 2nd 1924).

 

Even high voltage corona discharge methods were investigated for a short while but were plagued with problems. Likewise, the gold leaf Thermophone had some approval. Here a six inch square sheet of gold leaf was tensioned in a frame. A direct current was passed through the leaf creating some heat. When varying sound signal was superimposed 'modulated' ohmic heating varied the thermal expansion. The result was air movement creating sound. One other solution, with work beginning from the 1920's, was the electrostatic loudspeaker. The idea was relatively simple - two electrodes (one stiff, the other foil) separated by a dielectric created a capacitor. A force (Coulombs Law) is experienced by both electrodes if they are equally charged by similar or opposite potentials. If now the electrodes are modulated by audio signals, the thinner (foil) diaphragm will displace as the force between the two varies with the applied signal. The result is movement (overwhelmingly in the diaphragm electrode) creating sound.

Efficiency Vs Quality

 

Rice and Kellogg noted that the electrostatic speaker they had evaluated (electrodes of graphite painted felt, separated by sheets of varnished cambric) was capable of 'very fine quality reproduction' but required a very large surface area due to the small forces involved which were limited by the low dielectric breakdown of air. Furthermore, the efficiency was even less than that of a typical dynamic moving coil speaker. Nevertheless, a number of workers pursued this approach from the middle 1920's on - Colin Kyle in the US stands as one of the first to set the scene for innumerable developments up to recent times (Kyle C US patent No:-1,644,387 4th Oct 1926). Some 95 patents can be identified up to 1953. The best commercial example was undoubtedly the Quad (Acoustic Manufacturing Company) electrostatic speaker. Wonderful to hear but because of its generic capacitive effect, uniform drive, low displacement and stiffish diaphragm - true of most electrostatic units - better in the upper registers (and much better with an auxiliary moving coil bass unit).

 

Technology

 

As noted, early moving coil dynamic speakers were limited by crude construction techniques and less than ideal materials. Poor magnet technology meant that sensitivity was limited - indeed, as we have noted, all early examples had mains energised electro-magnets in place of permanent magnets (some producing dangerously high discharge voltages on switch off!). High intensity magnets were introduced in the late 1940's using highly ferromagnetic alloys (eg Alnico) but even so, the lack of synthetic materials with the right characteristics meant that pulp-based cone materials and voice coil surrounds were inadequate for any long term or predictable behaviour. Paper composite cones buckled as humidity changed, surrounds became stiff with age, voice coils would break away from the 'former' as drive powers increased. Attempts to improve matters ranged from very light weight titanium and polymer cones, to using silver wire voice coils. Cast chassis' were better than pressed steel, but casting costs meant that stiffer pressings were needed to match the advantages of cast chassis' in mass produced units. This was done, as were changes in adhesives and cone materials in order to create stable operation.

 

For all its problems the moving coil loudspeaker remains the primary method for sound reproduction and was really the only practicable technology. Over the decades improvements in speakers have come slowly. Outstanding in modern theory has been the electrical analogies and circuit modelling of speaker systems by Neville Thiele and Richard Small. Through them, and others, a better theoretical understanding of loudspeakers has been won by repeatedly refining the electrodynamic model involved. Because of this, recent times have seen major refinements in their use and construction - active motional feedback systems (typified by Philips designs) which have transducers and amplifiers to measure the cone displacement (and correct it), mean lower distortion levels in the bass register - but ever improving design methods now allow this to be realisable in standard units.

 

Ultra-high strength rare earth magnets are currently favoured for high stability, field strength and low mass/volume, but strangely in at least one instance (that of cast metal magnets) an earlier technology has risen again. Co/Se/Al/Fe (Alcomax 3) magnets have been introduced by one manufacturer to overcome the subjective degradation in sound caused through the introduction of anisotropic Barium Ferrite magnets. Though BaFe magnets had distinct advantages, intensive investigations produced the surprising recognition that cast metal magnets, being electrically conductive, were acting as a shorted turn around the voice coil thereby minimising flux modulation (thereby audibly decreasing sound colouration).

Similarly, the use of horns has appeared again in the form of acoustic lenses and labryinths to improve radiation efficiency. Likewise,. modern speaker materials make for high sensitivity with very low distortion (circa 1-2% Total Harmonic Distortion) commonplace. Pulp mesh for cones is carefully controlled, the surround edge needs to be 'significantly thinner and more flexible than the cone body'. Careful selection of the composite fibre orientation at the cone termination (near the outer surround) is important to fix the cone's exact compliance, damping and stability.

 

Twin coil and twin cone units are now ubiquitous, as are planar polystyrene and metallised cone bass units and very high power units employing ferrofluids (to transfer heat). High conductivity (silver wire) voice coils (to minimise heat generation) can be had - sometimes handling twenty times the power with much improved efficiency.

 

This is accompanied by a wider, flatter bandwidth for the same size speaker compared to fifty years ago. Yet, for all this, some argue that loudspeaker design is still a black art and that the perfect loudspeaker will forever elude us, one day perhaps, one day!

 

Notes

 (A force F, is experienced by a coil (and a magnet) equal to the magnet's standing field strength B, the coil length of wire I, the current magnitude i (determining H) and the angle between the coil and standing field - defined as Sin Theta. Thus the force F=Bli SinTheta (but where the angle is 90° then Sin 90°=1 and is discounted).

(2) After Dummer Electronic Inventions 1776 -1976.

 

Additional References

1. Encyclopaedia Britannica 14th Edition (1932) Articles: Radio and Loudspeaker.

2. 'Discussion on Loudspeakers' Jour I.E.E. March 1924 p 265 et seq.

3. Hanna & Slepian "The Function and Design of Horns for Loudspeakers" Trans AIEE 1924 p 393 et seq.

4. Round H J. "Moving Coil Horn Loudspeakers" Wireless World. Dec 17 1924.

5. Wente E C & Thurus A L "A High Efficiency Horn Type Loudspeaker" Bell Technical Journal. 1928 p 140 et seq.

6. Round H J "Broadcasting and Loudspeakers." Year Book of Wireless Telegraphy and Telephony 1925 p 825 et seq.

7. 'Loudspeakers' Volumes I, II, III, IV -anthologies of papers published by the Audio Engineering Society, PO Box 645, Slough, SL1 8BJ.

 

Acknowledgements

The author acknowledges with thanks the folllowing for helpful discussions or for supplying patent literature and other source material:-Mr. Paul Condon (Tannoy Loudspeakers Ltd), Dr, Peter Fryer (Research Division B&W Loudspeakers) Ms Ann Peacock (British Library), Ms S. van Dulken (British Library)  Ms Rachel Ford (Central Library, Birmingham) Mr. Xerxes Mazda (Science Museum, London), Ms Tracy Farrington (George Green Library, Universoty of Nottingham), Mr. Robert May (Clifton Library, Nottingham Trent University), 

Go to: Part Three

The Tannoy dual concentric 15" speaker - 40 years in the making to achieve a specification and operational performance second to none (Courtesy Tannoy Ltd)

Cutaway section of a modern ultra high quality speaker boasting excellent sensitivity, high power handling and very low distortion (Courtesy B&W Speakers Ltd)

The Rice Kellogg experimental annular field moving coil loudspeaker.  The 45 degree paper composite cone suspended on a free moving, sheet rubber surround

Early Electrostatic Speaker employing felt impregnated with graphite and seperated by 'two sheets of varnished cambric'

Gaumont Type Loudspeaker, 1924

Rice Kellogg 'Free Edged' Speaker with motor (voice) coil

View of Rice & Kellogg's Laboratory, circa 1926