OCR-excerpt from HifiNews
Original printed September 1981,
Reprinted Februari 1999

Quad ESL-63 by Trevor Atwell, Part 1
THE EMERGENCE of the ESL-63 must be the longest-anticipated surprise in hi-fi history, and certainly one of the most welcome. It follows some 18 years of development (hence the '63'). Meanwhile the 'old' original electrostatic loudspeaker, continuous in production after fully a quarter-century, is still a standard against which others may be judged, notwithstanding its slight imperfections. This would be a staggering achievement in any field, and particularly in the fast-changing one of hi-fi. Quad's policy of making a real advance with each new product, followed by very long production runs and service facilities extending over decades, is in sharp and welcome contrast to the facile trendiness with which so many makers offer this year's perfect product, only to displace it next year with a minimally tarted-up version presented as even more shatteringly 'perfect'.
Before getting down to the review proper, it may be useful to offer some simple technical background, since the electrostatic principle often gets short shrift in the textbooks. Odd, when one considers that the reciprocal device, a capacitor microphone (the scale is different!), is by far the most important professional type, and for good reasons. most of which are shared by the ESL.
I skipped the technical description, because the pictures are not included here, buy the magazine if you are interested (MT).
Electrically, the rate of doing work is given by the product of voltage and current, and in this case the voltage at any instant is that supplied by the transformer output winding, while the current is simply the electron flow between the plates. In a simple, perfect capacitor no work would be done. and the current would be 90 degrees out of phase with the voltage accordingly, but here the air motion introduces a loss, or acoustic resistance, to give an inphase, dissipative component.
In this very basic explanation we have used mainly the electrostatic potentials, but in practice it proves more rewarding to consider the current flow into such small increment of area that makes up the total, and relate these to the air pressures that result at any given point in the sound field. This may sound involved, but it isn't really, and it leads to the result - rather surprisingly - that the acoustic performance of the loudspeaker can be totally described in terms of these currents. Those interested in pursuing the argument further should consult the paper 'New Developments in Electrostatic toudspeakers' by P, J, Walker.J,A.E,S, 2S. p,795 (Nov, 1980).




The electrostatic principle has a number of advantages over conventional moving-coil drivers. The moving membrane can have a very low mass, so that very little mechanical energy is stored in it and hence it will exhibit little overhang. It can also be driven all over its surface, avoiding break-up modes which are the curse of the moving-coil unit. The large area couples effectively to the surrounding air, avoiding the need for an enclosure and so eliminating possible colorations from reflections within an enclosure as well as from resonances in the walls. The system also has an inherently better distortion performance.
Naturally there are also some snags. To start with, a large, flat radiator is not much use at high frequencies (which is why tweeters are small). The reason is the destructive interference (cancellation) between sound waves from every pair of points on the surface for which the difference in path length to the listener is an odd number of half-wavelengths. This gives a radiation pattern with a sharp forward lobe and a set of annular side lobes at various frequency-dependent angles. Only the narrow front lobe will provide reasonable stereo imaging -not very much use! In the original ESL this is improved by dividing the frequency range between two outer panels, which shared the lower frequencies, and a narrower central panel used for the high ones. The vertical dispersion is extended somewhat by curving the panels about a horizontal axis. These measures result in a difficult and rather labour-intensive construction, while the listening area available for good stereo performance is still limited.
Next, we cannot simply have a charged membrane in isolation, because there is always some leakage, and the theory of operation calls for a constant charge. The membrane is therefore coated with a conducting film of negligible mass and very high resistivity, fed from an internal EHT generator which is connected between the membrane and the centre-tap of the transformer winding. This keeps the membrane charged, with a rate of change slow compared with that of the lowest frequency to be reproduced. The EHT is about 6kV in the early model, and 5,25 kV in the ESL-63. Another problem is that a high voltage electrode makes a first-class dust collector - the principle is used in the electrostatic precipitators that reduce particle emission from industrial chimneys. So a dust cover must be introduced - it must be a good seal, but it must also be acoustically transparent without adding any rattles. A certain amount of damping is also needed, because regenerative damping is not practicable.




In the ESL-63 the worst of these problems - that of controlling the radiation pattern in a predictable manner independent of frequency - is solved in a highly ingenious way. The argument goes like this - an ideal source would be a single point radiating isotropically, for example, by pulsating in and out like a tiny balloon alternately inflated and deflated, giving spherically expanding soundwaves. So far, not very practicable - even if a source of this kind could be made, a sensible sound output would require an impossibly high amplitude at the surface. But, if we move away from the source, the total output power is spread over an area that increases with the square to the distance, giving an expanding spherical surface in which the air-particle displacement is correspondingly reducing. In fact, the neccessary displacement would be within reasonable limits at a spherical membrane M placed 300mm from the source. If such a membrane could be constructed and made to move in the same manner as the wavefront at that distance, then the actual source could be removed without anyone being the wiser, provided that the membrane was between them and the source. Such a membrane would be a stinker to devise, but suppose a plane one is placed with its centre coincident with that of M, which is then removed. The radiated sound has reached the centre, and will arrive at a circle of radius d1, after a time t1, radius d2 after time t2, and so on. It will be seen from the diagram that if the differences between successive delay times are equal, the radii of the corresponding circles will get progressively closer as we move out. If the membrane is now placed between two perforated plates on which are conductors arranged in concentric circles, and the signal to be reproduced is fed directly to a central circle, and also through successive steps of a delay line to the outer ones in turn, then the membrane movement will be that of a plane through which the soundfield is passing - exactly what is wanted. As indicated above, the required far-field can be realised specifically in terms of the currents flowing, including a correction for the limited size of a practical membrane. This is truly an elegant solution, and its correctness is proved by the fact that it works just as predicted.
Turning now to the construction and circuitry, there can be but few who do not yet know what the ESL-63 looks like - a brown cloth covered rectangular section 820 mm high x 620 mm wide x145 mm deep in the middle, tapering symmetrically down to 82 mm deep at the edges, giving a rounded front and back, sandwiched between two similarly shaped boards of real wood top and bottom, and standing on a hollow plastic plinth which extends behind the main frame. On the top of the rear consist ot four separate copper-coated and perforated boards of the type used for printed circuits. The top and bottom ones are completely coated, and the inner pair are divided into a central circle, five concentric rings and two outer symmetrical, but irregular, portions (all in two halves) by gaps etched out of the copper. The plates are about 5 mm apart, with the high-resistivity coated membrane supported midway between them. A sheet of a special cloth is inserted for damping purposes. A plastic dust cover, about 2 um  thick, surrounds the 'sandwich', and the cloth cover is supported on an expanded-metal frame. Inside the plinth are a number of integral ribs and pillars which support a surprising amount of electronics. Impedance transformation between the input and the transducer is by two large,  interconnected transformers at opposite ends. A mains transformer supplies the EHT with the aid of a Cockcroft-Walton multiplier, and also gives a low voltage which is bridge-rectified to feed protection circuits. The multiplier and protection circuits are on two small boards near the middle of the plinth. Directly below the main body two more boards hold the two six-stage lines that feed the sets of concentric rings on the back and front respectively. A small, separate board has a neon indicator, visible on removing a little plastic bung on the top of the plinth, whence it can be observed flashing while the loudspeaker is energised. This is purely a service aid, and there is no need to worry about how often it flashes, or whether both speakers flash at the same rate (they probably won't). The rule is - if it sounds all right then it is all right!




The protection circuits are particularly ingenious. In a conventional moving-coil loudspeaker any damage is normally due to overheating a voice coil, and the thermal time-constant involved allows plenty of time for a relay, driven by a circuit of greater or less sophistication according to price or maker, to remove the drive. In an electrostatic unit the damage is caused by an arc between the electrodes which burns a hole in the membrane. The arc is initiated following ionisation of the air, which occurs about three orders of magnitude faster than the operation of even a high-speed relay. Ouad get round the problem with their usual flair. The first part of the protection is a soft-limiter that comes in when the input exceeds some 40V peak, and does clever things at the input to the transformers without placing undue strain on the amplifier. This softclipping gives progressively more audible distortion, encouraging the operator to lower the wick. Should that person prove intransigent, ionisation will eventually begin, and will be detected by a small aerial built into the bottom of the electrode assembly. This signal is used to 'crowbar' the input for one or two tenths of a second, after which the circuit takes another look, and shuts down again if necessary. Amplifiers lacking short-circuit protection should not be used, therefore - most do incorporate it, but check with the supplier if in doubt. While the protection circuits are operating, power is being dissipated in resistors in the speaker. These can cope with normal music inputs, ie. intermittent overload, but continuous overloads must be avoided. Any steady-state test signal should not exceed 10V RMS.
In the event that an attempt is made to operate the ESL-63 with no mains input, and hence with the 'RF sniffer' circuit disabled, the audio input is made to crowbar itself at the safe level of about 8.5V RMS. This is essential, since the user cannot hear anything, and will tend to push up the gain to very high levels-we must all have carried out this illogical sequence at one time or another!
The first part of the listening tests involving a good deal of experiment with loudspeaker positioning. The ESL-63 is a doublet (hence the acronym FRED, or full-range electrostatic doublet) which radiates equally fore and aft, but not at all in the plane of the membrane. The basic polar diagram is therefore 'figure of 8', except that Ouad, after much experiment, deliberately narrow it with increasing frequency, maintaining the symmetry about the listening axis. The ESL should not, therefore, be used with a wall close behind it. Ouad point out that the optimum position for any loudspeaker is 1/3 of the way along the room diagonal, but acknowledge that this is seldom practical, suggesting a minimum of 600mm from the rear wall. The polar response is such that the lateral and vertical room modes will not be excited if the loudspeaker is at right angles to the side wall. However, angling towards the listener is recommended, allowing a degree of excitation of the lateral mode. As long as there is enough space, at the rear, some angling increases the path length of the rear radiation (and usually the number of reflections) before it eventually reaches the listener. In practice, the results from any doublet will be both room and position dependent, and there is no substitute for experiment to optimise performance. In my particular habitat the final position for each speaker was about 1.25 m from the end wall, with the nearer edge 200mm from the side wall, angled to face the listener seated a little over 2 m away.




Several owners have reported a little chestiness when used on the floor, which was removed by raising the units about six inches or so. Ouad say that this may occur in small rooms over shortened listening distances. My listening room is not small, but it does have a 'honky' floor Several owners have reported a little chestiness when used on the floor, which was removed by raising the units about six inches or so. Ouad say that this may occur in small rooms over shortened listening distances. My listening room is not small, but it does have a 'honky' floor of boards on joists with a small underfloor space, and I did notice the benefit of using blocks about 130 mm high under each speaker. Evidently this is also a matter for individual trial.
Having used the 'old' ESL for many years, and found nothing to beat it for open transparency,  particularly in the midband, I was not fully prepared for the degree of improvement achieved by the ESL-63, which actually makes its progenitor sound coloured! With the best input material one is first amazed and then delighted with the sheer realism of the sound, which places you where the microphones are, with the imperfections and size of the listening room as the only occasional intrusion. The stereo imaging is far better than anything else I have ever heard - the positional definition and image depth are astonishing. Unlike its older brother the image remains stable over a wide listening area, and the lack of any vertical or horizontal interference patterning as one moves one's head is unique and most refreshing. In a good source it is possible to locate, for example, individual players in an orchestra, and even to distinguish between two performers sharing the same desk! Compared with the earlier model the bass extension is considerably increased, with exceptional firmness, while a touch of chestiness in voices and a slight exaggeration in the presence region have both vanished. Transients are crisp, and the sheer detail in every sound is a revelation. Such superb performance makes a real advance in an area in which progress is now slow and expensive, and I find it hard to see how any multi-way movingcoil loudspeaker can be made to overtake the ESL-63, except in sheer output power. Indeed, a significant difficulty in carrying out listening tests on the ESL has been that of concentrating on sound quality instead of simply enjoying the music. What else does one want from a loudspeaker?
Naturally a reproducer of this quality shows up recording imperfections very clearly. There before you are the cardboard cut-out victims of the worst excesses of the multi-mike technique,  the violins huddled in a little box on the left, the brass likewise somewhere else, and a singer (apparently in a different building) sticking out in the middle. There, too, are the pipe organs 150 feet wide with pipes arranged and diffused in a way never dreamed of by the builder. Thank goodness some early ESLs went to recording studios, where those who wish to hear what they are really turning out can do so! Certainly record reviewing will never be the same again, now that the really good stand out as clearly as the really awful.
For those who want to hear sounds at realistic levels down to 20 Hz or below, which means only those with access to original tape recordings of pipe organs, or electronic sources, a sub-woofer will be needed, as with any other loudspeaker that would generally be regarded as domestically acceptable. Here the ESL-63 with an Audio-Pro B2-50 (or40) make an unbeatable combination - at a price!
So much for the quality - what about the volume? Significantly higher than the older version,  and plenty to satisfy not only the classical listener but also a large proportion of pop fans too.  For the really heavy stuff it will still be necessary to stick with the moving-coil, though, as it will if you prefer an 'exciting' (ie. unreal) sound.
This product is so important and so different in principle from the more familiar loudspeakers that it demands an unusually long review. Space limitations make it necessary to leave the measurements, which also call for some discussion, and the conclusion until the next issue.  Perhaps no great psychic powers are required to guess the general nature of the final summary,  but I will give those who only started at this paragraph a little hint - the review loudspeakers are being bought to act as standards against which other loudspeakers will be compared in future reviews. I should be delighted to find one that betters the ESL-63 in sound quality and imaging,  but I would suggest that if you're going to read a book while waiting, you'd better make it a thickish one!
Next month the measurements.