Bettering Quad Concentric Dispersion Rings with Digital Delays.

[owenhamburg]

Electrostatic speakers can have bass, if big enough, with solid supports, also Electrostatic speakers can have excellent treble and mid range (in my opinion unsurpassed). Electrostatics typically do not have good sound dispersion. Particularly the high frequencies decay very fast of axis. This is simple wave physics of a planar driver.

At this moment in time all Quad models since the ESL 63, including ESL 2805, ESL 2905, ESL 988, ESL 989 have used a concentric ring structure to improve dispersion.

I was wondering how much further this idea could be improved upon. The Quad uses about 9 separate rings. An analogue delay + filter circuit that is ingenious and keeps the costs down for an electrostatic with better dispersion.

With digital sources time delays are easier and more flexible. I propose rings less than 1 cm wide to the diameter of the speaker we might have a hope of making electrostatics have great off axis treble, or at least better, and give each ring nearly full audio range, since they wont be canceling each other off axis.

Many driver transformers, and many amplifiers, each driven with a separate different delayed digital signal, each ring could be given a accurate time based delay. The Central ring is defined as 0 time delay, driven full range, and each ring around the center given a time delay and a filter range dependent on radius and width of the ring respectively, and the position of the simulated point source. Volume balancing might also be used to compensate for the lack of high frequencies in the wider panels.

Will the Speakers concentric rings need to shield the listener from the simulated point source? If not then very wide sound stages could be generated.

While the central rings runs full range others are limited to their wave length relative to the width of the ring. My estimate that to get 4KHz imagine

A small treble unit with thin stators might be able to get significant off axis responses if mounted suitably and damped nicely behind.

Maybe filtering through analogue or digital methods before the power amplifier would prevent cancellation for some wider rings in the panel.

Maybe starting with a pair of Quad treble Stators or similar to this wonderful looking DIY design Elektrostaten Project 04 and 10 to 20 audio amplifiers and associated step up transformers.

Have any commercial or DIY electrostatic builders tried this?

[Calvin]

Hi,

I can´t share Your emphasis about the technique of delayed ring-segments, since it has serious drawbacks and similar or better results can be reached with less effort.
Since the effort regarding development and special parts count is high, this technique is not at all a lowcost solution.
A simple resistor segmentation solves the task of increased dispersion without the serious drawbacks of the ring-system.
These are:
- effectiveness of the ring system on dispersion is rather small. Simple curving or R-segmenting gives similar or better results
- area of each ring is too small for adaequate dynamic range
- small area results in incredible high impedance values
- impedance values ask for ridicolously high transformation factors
- reduced bandwidth and low overall quality of transformer behaviour
- the point sourve -which it should emulate- might be the theoretical ideal under freefield conditions, under real room conditions it most certainly is not. A cylindrical or dipolar cylindrical shape serves most room conditions better
- driving each segment through a dedicated delay, amp and transformer increases effort to the ultimate without gaining much -if anything(!)- in return.

jauu
Calvin

[kavermei]

Hi,

personally, I quite like the line source since it is so easy to get excellent soundstage imaging. But you pay the price of having a rather narrow sweet spot.

When you use rings on a plate stator, you lose a little stator area each time you create an extra ring, because there has to be a non-conducting gap between the rings. This limits the amount of rings you can have. In a wire stator design, such as the 'spider' by Tim Weert you mentioned, you don't have this problem.

Try to find the book chapter on the ESL63 by Baxandall. He explains in a lot of detail the ring radiator design of the Quad. IIRC, one of the conclusions was that there should have been more sections but it was not possible from a construction point of view.

Then there is of course the problem of needing all those transformers and amplifiers :s

Kenneth

[Capaciti]

Hi,

you might think the delay line could be improved regarding dispersion of high frequencies. Thats not true

The delay has nothing to do with dispersion but with timing. If you do some measurements one can recognize that step response is a little bit better than without delay. If you can hear it ? i doubt ?

The poor dispersion of the quads is caused by the diameters of the concentric rings. especially the inner ring for the high frequencies has too much diameter , which is needed to have sufficient radiation area. To improve dispersion at 10000 Hz and higher the ring need to have a diameter of 3cm only while it is more than 8 cm !!

Capaciti

[harry_at]

Originally Posted by Capaciti

The poor dispersion of the quads is caused by the diameters of the concentric rings. especially the inner ring for the high frequencies has too much diameter , which is needed to have sufficient radiation area. To improve dispersion at 10000 Hz and higher the ring need to have a diameter of 3cm only while it is more than 8 cm !!

Hi Capaciti,

with a change of the concentric rings to vertical lines like in segmented wire stator, the diameter of the new high frequency section would be much smaller and better suited for dispersion. Is such a change possible or would it counteract Peter Walkers idea too much? A simulated line source.

Originally Posted by Capaciti

The delay has nothing to do with dispersion but with timing. If you do some measurements one can recognize that step response is a little bit better than without delay.

As Calvin has mentioned a resistor segmented design would be much simpler. But your comment about timing is interesting. Maybe this is the reason why this speaker has something special.

Harry

[Calvin]

Hi,

Is such a change possible?

Yes, of course. The old AudioExklusiv P3 and P6 worked with delayed strip segments.

jauu
Calvin

[LineArray]

Originally Posted by Capaciti

Hi,

To improve dispersion at 10000 Hz and higher the ring need to have a diameter of 3cm only while it is more than 8 cm !!

Capaciti

Has someone tried an acoustic lens e.g. made of foam in front of
the inner "disc" ?

[moray james]

I modified a set of Acoustat 0ne plus 0nes with vertical layers of light damping material both front and back many years ago. I left a central strip of about one inch open with no damping material then I progressively layers the material in three sections outward with each section having one more layer than the previous one. The results were good but it was a fair amount of work. I used light interface material made from non woven polyester which is used to give extra body to clothing made from light weight material. I think that using 1/16 inch thick pure wool hobby felt would probably have done a better job but never tried a second time. So yes the idea works and the panels were well controlled as I never had any stability problems with the panel.

[LineArray]

Do you mean "stability" in imaging ?

[moray james]

Stereo image and sound stage were a little better with the sweet spot opening up some. I found the speakers a little easier to set up not being quite so fussy to position. It is as if the panel was narrower since it only radiated full high frequency over a one inch width in the centre of the panel. The other thing that I tried which helped was to place a deflector (or beam splitter) behind the speaker which made positioning easier again and also helped to improve the image some. I used a length of plastic plumbing pipe about a four inch diametre. I think that a larger diametre pipe would have worked even better say 6 inches but I did not try that. Charlies beam splitter does exactly the same thing.