Design Philosophy ESL III Electrostatic Loudspeaker
Whilst the ESL III loudspeaker has been designed primarily for the home builder, there have been no compromises made to performance or sound quality to achieve this aim.
Indeed, the performance of the ESL III is up amongst the best of the commercially available electrostatics at a fraction of the cost.
The three flat panels that make up one loudspeaker are arranged in a curvilinear array ie. the bass panels, which are placed on either side of the centre treble panel, face slightly outward of the speaker centreline.
The outer panels (called bass panels) are actually full range but the high frequencies have poor dispersion due to the width of the panel. To compensate for this the centre treble panel is narrow giving good dispersion which sonically “fills in the gap” between the two bass panels.
The bass panels have an array of node points running through the vertical axis, which are spaced in an asymmetric array but are mirrored from the speaker horizontal centreline. This creates a sort of D’Appolito array where the diaphragm is divided into sections of unequal sizes from the horizontal centreline in up and down directions.
By distributing the resonances across a range of frequencies some compensation can be made for output losses caused by cancellation effects. This also stops adjacent sections of diaphragm having the same resonant frequencies thereby giving a dominant resonance, or “one note bass”. This approach also reduces anti-node formation at certain frequencies, which result in “dead” or redundant sections of diaphragm.
The result is a reasonably wide sweet spot (always difficult with panels) where you don’t need your head in a vise to maintain the excellent soundstage, plus a reasonably flat, extended frequency response, which does not need signal modifying (destroying) capacitors or inductors to shape it to be acceptable.
To enhance the performance further it was considered essential that this should be a zero shunt capacitance design.
As an example, ESL panels, which have air gap spacers that are simply glued onto an active grid, may be simple to build but there is a serious trade-off. The area of grid that has the spacer covering it is non productive - it cannot drive the diaphragm, BUT the amplifier is still driving the capacitance it presents.
This means that on one speaker the size of the ESL III there would be over 1200 square centimetres of grid doing no work but still presenting a load to the amplifier. This means that the speaker has to be larger to achieve the same output, which then increases the capacitance again.
The method adopted in the ESL III is for the grid to be supported away from the air gap spacers on a non- conductive plastic matrix. This means that all of the grid is contributing to the output. The grid design is also important in ensuring planar movement of the diaphragm. Our grids have a solid outer border designed to increase the electrostatic field at the suspended edge. This reduces the length of the curve in the diaphragm that naturally starts to occur with the greater spring tension that exists close to the suspended edge.
As the grids do not extend to the outer edge of the panel, there is an added benefit from this approach by the elimination of arcing or tracking between the two grids on the outside of the completed panel, a notorious place for this fault to develop.
Grid (stator) design
As a support structure was to be used to reduce shunt capacitance, this gave the opportunity to use the thinnest but stiff material available for the grids.
The gauge of steel that is used is only 0.6mm thick.
This has a profound sonic effect.
Whilst using grids of thicker material does not cause high frequency roll off to any serious degree (that is until they get to around 2mm thick), they are unable to retrieve low level detail completely. This is due to the air mass that is contained in the individual holes. Effectively the holes form a small resonator, which destroys low level information. The frequencies at which this occurs, are linked to the ratio of the thickness of the material versus the diameter of the hole and is too complex to thoroughly explore here (even if I understood it fully).
However, it has been found that thin grids simply sound better so we adopted this approach.
This, needless to say introduced another problem, how to insulate them effectively. Being thin and perforated, all insulating materials tended to migrate away from the edges and insides of the perforations leading to flashover and leakage problems.
After many months of research, trial and error, we had a special powder coating produced, which overcame these problems. This helps enormously in producing a reliable electrostatic speaker.
Thin grids tend to “ring” when producing sound so it was important to provide some damping to the grid to deaden the colouration that would occur. This is especially important on the newer “windowed” support matrix, as there is a large area of unsupported grid material.
We found a polyurethane adhesive, which not only bonds the components together very well, but provides a superb damping medium for the grids.
This adhesive is used throughout the construction of the ESL III.
The ESL III was originally released some years ago using a heat-shrink polypropylene material. This gave very good results sonically, was easy to apply as it was self-adhesive (heat activated), heat shrinkable to the correct tension and has a very high dielectric strength - it would never arc through.
The problem with it was its weight. It was only available (and still is) in 25 micron gauge. This caused it to be a bit “mellow” in its sound presentation. The transients were not quite as good as they should be, although much better than most box speakers of the day. This was noticeable when compared A to B with much more expensive commercial units. To improve this, we started experimentation with thinner films eventually settling on a 3.8 micron polyester film. (6 times thinner than the original film). A technique was developed to enable the home builder to use this film with consistent, repeatable results.
The conductive coating we use was developed by us and is proprietary to E R Audio.
It is not an ionic solution such as static dissipants and does not rely upon ambient moisture to make it conductive. Instead it is Carbon based (no secret here, it’s very black) which is suspended in a high dielectric material.
It is humidity independent meaning that the weather will not affect its performance. We have tested it extensively in climatic extremes, from the humidity of coastal China to the dry of the Australian desert regions.
The surface resistivity is greater than 2000 megohms per square when fully cured. This means that charge migration will not occur, even at the lowest frequencies.
This gives good bass performance and reliability.
The audio transformers we use are another proprietary design of 1:90 turns ratio.
These are a multi section, interleaved design with a continuous power handling capability of well over 100 watts.
Low distortion and faithful reproduction of signal sources such as square waves was a priority.
As these were specifically designed for electrostatics with very high voltages envisaged, a second priority was placed on internal insulation and high voltage enamel for the winding wire.
The turns ratio of 1:90 gives sufficient drive voltage, even from lower output amplifiers, without putting too high a demand on current.
However, amplifiers with good current capabilities are necessary for high sound pressure level listening.