MartinLogan CLS loudspeaker Martin Colloms, January 1987

Martin Colloms wrote about the CLS in January 1987 (Vol.10 No.1):

Panel speakers are back in fashion. After years of determined effort on the part of a few established manufacturers, the number of these designs has seen a marked increase, and many new exotic models have appeared. Stereophile has recently reviewed two leading examples of the genre, the $2780/pair Apogee Duetta and the $2490/pair MartinLogan CLS, (both in Vol.9 No.7). To judge by the tone of letters arriving at the magazine's offices, the reviews generated heated controversy. John Atkinson asked me, therefore, to conduct an in-depth examination of these two models, to give a fuller picture of what these speakers are capable of.

I hope to present rather more data than contained in a conventional review. To begin with, there will be new supporting evidence on the sound of the CLS generated in alternative rooms. Furthermore, the review samples were of recent vintage, with the Logan equipped with the latest electronics. The CLS frequency balance had also been changed for the UK market. This involved the substantiation of a higher resistor value in the treble section of the equalizing/matching network to provide a richer, and in my view, more accurately balanced sound. (Those comparing my remarks with JA's in Vol.9 No.7 should bear in mind that JA's samples had not been so modified.)

Design Compromises
The MartinLogan CLS fullrange electrostatic is an open-panel speaker and therefore must suffer the low-frequency rolloff limitation of a finite baffle. It copes with this by means of a designed resonance at 50Hz. The electrostatic principle is attractive in many respects, not least because it allows the use of an extremely light diaphragm—so light, in fact, that its vibratory contact with the air molecules may be used as a palliative, controlling, damping mechanism. Light diaphragms can move quickly, promising good high-frequency response, while their low inertia assures a low level of stored energy, promising an accurate transient response.

In principle, the electrostatic driver is very efficient in its conversion of moving electric charge into acoustic power. Problems arise, however, in interfacing the necessarily large electrostatic element to the room, as well as to the power amplifier. The science of electrostatics is the science of high voltages, with the polarizing field established by as much as 7kV applied to the inner shielded diaphragm. Ideally, the fixed-mesh electrodes on either side of the diaphragm need to be driven in push-pull at upward of 500V, and require isolation from the user. (In the case of the Quad, the electrodes are protected behind grounded mesh screens.) A thick, tough, black plastic insulating coating is applied to the CLS electrodes; these simultaneously form the external grilles of the speaker.

High-ratio step-up transformers need to be used to couple the amplifier to the system, while the input characteristic is not one which allows for optimum power transfer. An electrostatic offers an input characteristic which is predominantly capacitive, hardly an ideal load. By the time these considerations have been taken into account, as well as the need to improve the poor dispersion of a single large diaphragm, the much-vaunted efficiency has been whittled away. Nevertheless, most electrostatics have a basic sensitivity of around 85dB/W (8 ohms), which is rather better than that achieved by present large ribbon systems.

The CLS diaphragm is unusual, for although it is only 0.004" thick, and consequently very light (the whole system is reckoned to weigh the same as one cubic inch or 16.4ml of air), it is formed from a pretty rigid plastic and is almost self-supporting. No mechanical damping is used in the diaphragm; its stretched "skin" and "plate" vibration modes are partially controlled in the first instance at the boundaries, where a lossy foam-plastic mounting is used, and by the acoustic impedance of the air load imposed on the diaphragm. For comparison, the Quad ESL-63 uses an 0.00137" diaphragm, with 0.001" plastic film for the dust-cover membranes. The CLS uses a density of Mylar similar to Quad, giving a total mass of around 0.02gm. (The ESL-63 is even more extraordinary, its diaphragm weighing just 0.003gm, or 3 milligrams, not too far removed from the tip mass of a pickup cartridge!) Another interesting comparison is with a typical wide-band moving-coil dome tweeter, where the moving mass is normally around 200mg. Quads enjoy higher levels of air-damping than MartinLogans and, in addition, use an interlayer of the sheerest gauze to provide necessary resistive mechanical contact damping to the diaphragm surface.

From its appearance it is obvious that the CLS diaphragm is broken up into discrete radiating areas by damping strips. It is possible to acoustically excite these separate "cells" by blowing on them. As when air is blown over the necks of differently sized bottles, each one has its own distinct "sound." By this means, dominant resonances are moderated and dispersed by dimensioning each rectangular cell differently (rectangular, in any case, to reduce the resonance-mode symmetry).

The vertical side compartments of the CLS diaphragm are of fairly high Q, and are tuned to 50Hz. Above resonance, this level falls at a natural rate of 6dB/octave, partly compensating for the naturally rising 6dB/octave response of the panel as a whole. Beyond this frequency range, the double-section, frequency-compensated, step-up transformer applies a further reduction in mid/treble level. Finally, the distribution of the cells and the arc of the curvilinear diaphragm complete the frequency balancing.

A full-range planar diaphragm like the CLS becomes increasingly directional as the wavelength of sound approaches the size of the panel. In the Magneplanars and Apogees, this is solved by allocating the subsequent frequency range to narrower and narrower line-source elements. For the Quad, the diameter is subdivided no less than eight times, providing a controlled reduction of element size with frequency, and thus maintaining a good radiation angle. MartinLogan deals with the problem in a different way: the electrostatic panel is formed into a near-parabolic contour with a prime radiation approaching a vertical half-cylinder. With this broader-angled distribution, some of the natural rise in frequency response is also offset.