Quad ESL 63 - A Short History

Details
Hits: 12047

Development of the Quad ESL '63 - A Short Patent History

The development of the speaker that became known as the Quad ESL '63 is largely the story of a search for greater power handling and greater output plus linearity at the low frequency extremes. It's designation as the '63 is a matter of conjecture, but most people accept as fact that 1963 was the year in which Peter Walker made his first notebook entries about the design that became the '63. It was a long development phase - 18 years in all, but as Walker himself said in 1994:

"It's like the ESL 63 loudspeaker. It took us 18 years to develop but it wasn't 18 years every day. [Laughs] Not at all."

Walker admits to leaning rather heavily on the work of Shorter from the BBC and Kellogg in his development of the ESL 63. He is plain about this in his 1979 AES lecture which was about the first occasion that the '63 had a public outing. In that lecture, he says:

"So, finally, what is new about all this? Really, it is a lot of old ideas fitted together. Kellogg in 1929 proposed the connection of a series of electrostatic elements by inductors as a delay line. His idea was to improve efficiency and reduce the power requirement from amplifiers. Shorter of the BBC took out a patent in 1941 describing the connection of a series of annular rings using resistors and Janszen, in 1953, suggested variations on the same theme.

In effect therefore, all I have done is to collect these ideas and add a little work which says that if you can make the device acoustically transparent, then the performance can be predicted. We think this is very important since it enables correction to the performance to be made very easily and after simple laboratory measurements." Peter Walker, June, 1979 AES British Section.

Oddly, Walker does not actually reference Janszen in the Patent on the ESL '63. The "suggestions" must have been fairly vague ones.

The work of Donovan Ernest Lea Shorter is particularly interesting. you may like to take a look at Shorter's Patent (G.B. Pat. 537,931) diagrams of July 14, 1941, before reading further. OK, so it's connected to a valve driven circuit. However, you see the basis of the delay lines in the ESL '63. Shorter was concerned partly in his work with directivity, and states:

"The present invention provides means for bringing about the necessary reduction in source dimensions with rising frequency, and is applicable both to those loudspeakers in which the radiating surface is divided into mechanically independent sections and to those in which there is used a single diaphragm having sufficient internal or external damping to attenuate substantially any transverse wave motion at high frequencies which may be propagated from one part of its surface to another."

The matter of transverse waves upsetting the apple-cart (as it were) is dealt with in the ESL '63 by gently electrically attenuating the signal as it propagates outwards from the central node. Directivity control is, of course, one of the things which Walker was trying to achieve in the '63, and to a large extent succeeded.

Another advantage of the approach is that it makes the device a far tamer load for amplifiers to drive. The original Quad, as we know, was about a 30W speaker at low frequencies dropping to below 3W at very high frequencies. The impedance curve is a sinuous one, to say the least. Efficiency and "tameness" are not pre-requisite criteria for high-fidelity reproduction however. It achieves this as follows:

"An incidental advantage of devices according to this invention is that, since the active electrode surface and hence the capacity, is reduced with rising frequency, the input reactance of the whole device can be kept more nearly constant than with a fixed electrode area. [Donovan Shorter]"

Furthermore, that old, often-wished-for idea of a rigid driving piston for the air in the room can be more nearly realised using Shorter's techniques:

"By distributing the force in this way it is possible to modify the form taken by the vibrating diaphragm surface at low frequencies from a curved contour with maximum amplitude at the centre to something approaching a plane surface over most of its area. Thus, for a given maximum excursion - the factor which usually places a limit to the acoustic power which the loudspeaker can deliver - an increased volume displacement, and thus an increased acoustic output is obtained."

So, you can push more air with this type of design simply because the diaphragm stays flatter, over a wider frequency range, and you can drive it harder without sucking the diaphragm into a stator ring. This is why the ESL 63 plays louder and lower than the original Quad ESL.

This is frighteningly close to what we see in the ESL '63, and we are talking about work done in 1941!!! No diaphragm materials that you could make it with then, and Walker seems to be the only person to have picked it up and read it later on. However, this is not the '63, as Shorter proposes that high frequencies be fed into the centre and that lower frequencies be fed to the outer rings only. It is quite plain though, from even a brief reading of Shorter's works in this one patent that the ESL '63 had its genesis conceptually in many ways, back in 1941.

Now, the work of Edward Washburn Kellogg pops up again. Peter Walker seems to have been rather a fan of Kellogg, and not without good reason. He referred to Kellogg's work when developing the original ESL too. The later Patent (U.S. 1,983,377) that Kellogg was granted is referred to again for Walker's work on the ESL '63.

As we saw in the history of the original Quad Electrostatic loudspeaker, Kellogg's invention also had the benefit of reducing the destructiveness of any sparking that occurred by reducing the amount of current that could be discharged through "adjacent coils". A short perusal of Kellogg's September 1929 patent clearly shows the inductors arranged between the individual elements of the speaker (in section drawing), much as in the ESL '63. He even gives a diagram of a single-ended embodiment of this type of device.

Between the two patents mentioned above you have the crux of the ideas that went into the ESL '63 - Walker admitted as much in 1979. However, he could not have obtained his own patent by simply copying the work of Shorter, although the work may have been out of patent by then, since there is no evidence that anyone renewed the thing. All the rest of the patents referred to are listed below, and they all deal with the other, less obvious aspect of Walker's additional \93little work\94 as he humbly described it. This is the aspect of Constant Current Drive and Feed Back. Yes, the ESL '63 uses a form of feedback to ensure that the directivity of the speaker is preserved. This is the unique feature that enables the ESL '63 to model the imaginary point source of sound about 30 to 35cm behind the plane of the speakers. If you care to look at the principal of reciprocity in another way, applied to this speaker, and connect the annular rings in reverse order, you can make the radiating source appear to be the same distance in front of the plane of speakers! For what that might be worth?

Anyway, in G.B. Patent 1,228,775 for the ESL '63 Walker clearly spells it out:

"A convenient method of controlling the vector sum of the currents in the individual sections (i.e. the total current at the input terminals of the network) is by means of a resistor 7 which samples the current flowing to the electrodes of all the sections and develops a feedback voltage that may be applied via line 8 as feedback to the amplifier that feeds the loudspeaker through the usual transformer shown at 9, thereby to render the total current proportional to the input audio signal."

He further states:

"In a practical loudspeaker, virtually complete and automatic control of axis response may be obtained from about 100Hz to 10,000 Hz. The high frequency limit is due to stray capacity and lack of acoustic transparency."

 

Referenced Patents for U.S. Patent 3,773,984

(chronological order)

All of the above seem to have been searched to find out if Walker's method of "constant current drive" was already in use somewhere.They are all concerned, in the main, with methods of low distortion feedback, in some cases (e.g. William Harry) the patent is about using feedback to control the diaphragm of a microphone, but the same basic principles apply. The two patents of Wilkins et alia, are about control of amplifier source resistance, and current-proportional feedback which was critical to establishing the \93vector sum of currents\94 that Walker refers to in his patent. The more you read these, the more you see the famous Quad equation Z1/Z2 = Z3/Z4, and things which are very much like it.

All in all a superlative piece of work also. In its "factory" form, many say the ESL '63 is not as "pure" in the mid-range as the earlier speaker. It can be modified to perform a little better, certainly in the bass spectrum. The purity of the high frequencies and mid-range may well be affected by the matter of that \93grille\94 that is internal to the speaker with its many little shallow lattice openings acting as individual small resonators.

In the end though, this speaker, like its hallowed predecessor has endured, in spite of fashion. It is a marque of quality and clear thinking.