The blue Claus was my first ESL design based on curved punched metal sheet stators and served as a case study to proof the concept.
The curvature of the stators raisens mechanical stability considerably.
A slight widening of the distribution character can also been experienced.
The sweet spot -or better the sweet line) remains still small.
Building effort and cost are higher than for a wire stator ESL-panel.
It also requires much better and costier audio transformers because one cannot segment the panel electronically and so the transformer needs to be designed for the full panel capacitance. Amplitude correction measurements need to be external, e.g. part of the crossover circuit.
Curved stators only allow for small diphragm excursions. They discard for fullrange panels.
So many good reasons speak against curved metal sheet stators.
As a pro they offer better efficiency and -depending on panel size- considerably lower transformation factors of 1:50 to 1:70.
The ESL-Mantra spells: Efficiency, Efficiency and Efficiency again.
It is a common beginners mistake (and all to often can be seen with professionals) to disregard this important factor and to try to create a -already terribly inefficient- fullrange ESL on first try.
With every Hz below 150Hz problems rise. Low frequencies demand increased stator-stator distances, softer diaphragm suspension, higher driving voltages, thicker insulative layers, higher transformation factors, more power, etc, etc.
The disadvantages are so serious, that the postulation of maximum efficiency becomes the most prominent.
Higher transformation factors always result in less good performance if everything else is kept equal.
Low transformation factors are as such a very important factor in the hunt for the finest sonic details.
Soundwise it shows in a more lively, more dynamic and more authentic playing.
Inherently inefficient ESLs like the Audiostatic, the Final -they restarted under the label PIO with the same construction principle after Final closed down because of too high numbers of dying panels- and as most prominent example the Quad, which sound terribly lame and dull in comparison.
Curved metal sheet stators require elaborate insulation procedures, while wires can be sourced well insulated off of the shelf.
It is especially the one coarse side of the sheets surface which creates problems to coat a sufficiently thick and safe insulative layer.
Unvisible from the outside this makes a great difference in performance and is a point of proof of the manufacturers KnowHow. It also shows in measurement values and sound.
As DIYer one is quite able to suffer alot. As such endless hours of preparation of coating and the procedure of coating itself were put into the metal sheets. The result is a durable and safe insulating layer, that allows for above average efficiency of the panel.
While typical (commercial) wire-stator panels run on polarizing voltages of 5kV-10kV and similarly high signal voltages these panels allow for up to 10dB higher SPLs with signal voltages below 2kV. Besides the sheer safety aspect lower voltages mean lower aging of all materials.
As audio trannies I used pairs of simple toroid power transformers (80VA) with single primary and single secondary windings of 230V and 6V.
The 6V-windings were connected in parallel and serve as primary, the original 230V primaries ar connnected in series as tapped secondaries.
The rather low inductance of the trannies doesn´t permit low frequency useage, but it is more than sufficient if crossover frequencies of >100Hz are at game. The frequency restrictions are sensitive in any way.
The flashover treshold from prim-to-sec is specified with >4kVrms. Thats much higher and promises a longer lifetime than with ´special´ audio-trannies. The coupling between the windings is very tight, which one can read from the large electrical phase values.
The upper bandwidth limit with the panel connected is typically alot higher, because a toroid has lower stray inductance compared to a classical EI-transformer. It is sufficiently low to supply for panels with up to 1nF of capacitance.
The inter-winding capacitances are higher. But since these enter the load-equation just 1:1 (contrary to the panels capacitance which enters the eqation with 1:U²) they play a negligible role.
The transformation-factor U calculates here as: 2x(230V/6V)xefficiency of the toroid (~90%) --> 1:70
A simple Cockoft-Walton cascade generates the polarizing voltage.
For maximum efficiency the ESLs diaphragm should be mechanically tensioned as much as possible. This of course leads to high values of Fs (>100Hz, base resonance frequency).
The working frequency range starts from approximately 200Hz. I chose 250Hz.
The Bass tower must orientate around this limit. It was one of the prime goals to generate a similar distribution character to the ESL panel plusminus one octave around this crossover point and the amplitude response should show a mirror-like response.
After theory the transition between Bass and ESL should then be seamless, non-audible.
The panel exhibits a dipolar-cylindrical character. It logically follows that the Bass needs to be a high Dipole-Bass tower. A Dipole is an open baffle system, that hardly changes the free-air parameters of the drivers.
But depending on the style of the casing the free-air resonance actually sinks. The Quality-factor Qt nearly remains constant. Interesting that the parameters change more the smaller the dimensions of the casing are.
In the end I chose a slightly trapezoidal design, of which the sides taper towards the backside (A-style). The dimensions were 8" wide and 10" high/bass-driver and 8" depth, which lowered the freeair resonance by 5Hz.
At a specced lower bandwidth limit of 35Hz the driver should feature a Fs of ~40Hz, ideally in togetherness with a Qt-value between 0.5-0.7.
Also the outer diameter was limited to 8".
The Westra KW180-2694 was the candidate I finally chose. With its diameter of 7.4", a Fs of 42Hz and a high Qt it fitted all requirements and costing less than 20€ a piece even 8pcs per bass tower wouldn´t tear a too big hole in the wallet.
The driver features a nicely coated paper diaphragm with a soft rubber surround, mounted in a steel basket with nice frontal optics and a shielded double magnet system. The parameters variance was acceptably low within 10% of tolerance. Able of +-4mm of linear excursion the Westra-8-pack can push enough air volume for high bass SPLs.
The dipole casing indeed reduced the freeair resonance down to 35Hz.
The amplitude response had the expected dipole-typical response with a constant increase in amplitude above 35Hz until the first chamber resonance at roughly 300Hz and a steep rolloff above, nearly a mirror-image of the panel´s response curve.
The amplitude increase above 35Hz is well behaved because of the high and close to ideal Qt of ~0.6 of the drivers and asks for just a bit of equalization. This allowed for a rather simple crossover which basically just supports the drivers own behaviour. There´s no need to push the drivers with excessive amounts of eq and amplifier power and they thank it with good sound.
The inevitable large height of the bass tower now allowed to place an electronic casing below the panel to bring both parts on the same level of height. This way the ESL can be listened to sitting as well as standing.
To reduce the effects of acoustic phase cancellation and to connect the panel with its electronic casing a special drawn aluminum profile came to use. The profile has a banana-shape crosssection and a u-style extension at one end, which can hold the panel. The hollow profile may be filled with sand or similar high damping material. The fitting end plugs to close the profile came as accesoire with the profile. A pair of curved aluminum braces of rectangular shape are all thats needed to give the structure strength and stability using jsut 4-6 screws.
The sonic result was more than just pleasing. The theory proofed in praxis in a very impressive way.
Heavy paper and light film can play in true harmony.
The ´speed´ and precision of the bass was remarkable and on par with the ESL-panel.
The sonic character of a dipole bass is quite similar to the natural and clean impression of the panel.
Before, a couple of serious room modes and resonances had made it rather impossible to listen to deep bass of strings and pizzikati or an bassdrum. The dipolar cylindrical wave only exites one single room mode in axial direction. Precise, dry, with loads of contour and a crisp punchy kickbass one might just miss every now and then on the usual lowfrequency slam.
Only a dedicated subwoofer, restricted to a range below 40Hz-50Hz (below the room modes), could improve matters.
The possible dynamic range is alot larger than what one might possibly allow ones ears to listen to.
The very low distortion levels and the linear increase in dynamics give the impression of unlimited reserves and allow to listen to the system without stress and fatigue over long time periods.
Well, in the end it makes a very remarkable difference if a small 1" dome tweeter -possibly crossed over at low frequencies- desperately tries to shrill a hard rimshot, or if a 465sqrinch-membrane tosses the same easily off the cuff.