Yes the Quad ESL-63, 988 and 989 all use several high-turns-count air core inductors as series elements in a ladder arrangement, with cross-coupled shunt caps to form a tapped delay line, also called a lumped element delay line. The ESL stator rings are tapped along the way, each concentric ring farther out getting a more delayed signal. Hard to describe without a picture. There are websites that show it. Try Googling the Quad ESL-63.
The length of the wire is not what causes the delay! The delay is created by the interaction of the inductance of the coils and the capacitance in the delay lines/capacitance of the speaker segments.
Quad started putting delay lines in the ESL-63 and continues with the 988 and 989. The delay line serves three purposes. It delays the start of the waveform to successive segments of the drivers in order to simulate a "pulsating sphere". I think it also "equalizes" (i.e. flattens the frequency response) of the speaker by reducing high frequency output relative to low frequencies. Essentially, the highest frequencies are applied only to the center segment of the speaker. Reducing the radiation area reduces the output level and increases high frequency dispersion in the listening room.
I assume your stators are made of segments of vertical wires. How are you implementing the delays - resistors or inductors? Do you have any results yet? I would be interested in trying something similar in the future.
No, most recently I've been toying with plastic grating with stainless woven material for the stators. I've built tall narrow drivers, but the delay elements have been only paper and very limited "breadboard" tinkering so far. I have simulated using both resistive and inductive elements combined (you really have no other choice in the real world when using inductors). The resistive approach is not new of course; the variable-width Acoustat designs of Jim Strickland come to mind. Even Quad incorporated a resistance element (shorted turn) in its inductors. I haven't decided on where I’m going with delays yet. One advantage is the potential to improve the impedance presented to the driving source when using R-L delays elements.
If I understand you correctly, you have multiple adjacent panels. I have considered doing the same thing, but doesn't this approach severely limit your low frequency response? I would think that the resonant frequency of each panel will be quite high. Also, maximum excursion will presumably by a bit lower than a larger panel(?). The total surface area may be large enough but does it translate to any meaningful mid/bass frequencies?
There is one commercial design using vertical aluminium rods as stator. Lowpass/delay is achieved by driving them via resistors that act as first-order LPF together with the intrinsic capacity between the opposing stators.
Bret, I've toyed with notions of conjoined vertical panels, each adjacent panel delayed, and also panels divided into smaller vertical stirips, each adjacent strip delayed. As for bass, I'm looking into a non-dipolar design with a few other tricks in mind. I'm not nearly ready to claim victory with this set of new design ideas, so I'm going to say little more until I have something that works enough to tell about.
Charles, which commercial design with the aluminum rods and RC "delays" are you referring to?
Well not measured...but, according to Peter Baxendal in Loudspeaker and Headphone Handbook the inductors has a nominal value of 3H. Further he says the lattice cap and the cap in the coil combined is about 132pF giving the delay line a characteristic impedance of 300kohm
Then "the air-cored inducters are sufficiently close spaced for mutual inductence between them to be very significant..."
Also pointed out the coils has a few shorted turns to damp the delay line.
Last I would recomend reading the chapter on electrostats in this book (isbn 0
240 513711) if you are interested in quad electrostats 57 and 63. Most theory so if you want building instuctions read another book.
Now if someone could measure the resistance in the coils..I think it can be done just measuring on the pcb.