Another ESL speaker project

[Jazzman53]

Hi all,

Since October of last year I've been toiling away on my biggest-ever ESL speaker project, and I'm very happy to be nearing completion after 5 months of exhausting work.

Several of my Carver Audio friends have twisted my arm for years to build speakers for them, but until now I've declined because all of my designs are exhausting builds, especially for an old guy like me. My friends prefer my lightweight Audi speaker because it's prettier and easier to transport than my other designs. I must have bumped my head somewhere because I finally agreed to do it.

I'm building four (4) pairs (8 speakers) of an OB hybrid design with a 12" woofer mounted on a semi H-baffle. The frames are red oak and the stat panels are electrically segmented wire stator type, which function as a line source projecting a cylindrical wavefront. Bending the wave front is accomplished using a frequency/phased-array of (15) wire groups driving the diaphragm sequentially from the centerline outward.

The stator wires are 20 AWG, UL-1061 SRPVC-insulated, single-strand copper. The HV bias supplies are simple 3.2kV diode/capacitor cascades, and the step-up transformers are tandem pairs of 50VA 230V/2x6V toroids wired for 76:1 step-up ratio. The diaphragms are 6-micron Mylar C with a 2-micron Licron Crystal ESD coating giving E6-E9 Ohms resistance. The speaker are bi-amplified using a DSP crossover.

Below are three videos showing the stator/panel build process and a completed speaker, along with a few build photos.

Enjoy!

Wire stator build video:

Panel assembly video:

Finished speaker video:

 

[oldsysop]

Beautiful !!!

 

[RayDunzl]

Make me a pair, too.

Very nice.

 

[RayDunzl]

Assuming that is a recording of the speaker, since I see the diaphragm fluttering...

Looks good.

 

[Blumlein 88]

Send it to Amir for Klippel testing!!!!!

That is a great looking result. You should be proud.

Just an FYI, I know later Soundlabs use an off the shelf adjustable high voltage supply. I've been planning for quite a while to change mine to that as the supply has weakened in one of my speakers. You might find it useful in future builds or upgrades.

I believe it was the CS2091 series HV modules made by HV Power Solutions Inc. Not sure that is available now, but there are similar HV units from various companies.

EDIT to add:
Found this page I'd seen before with some info on off the shelf HV supply. Mentions Emco units in the upper part of this page, and has more detail further down the page.

https://drmrehorst.blogspot.com/2018/10/diy-electrostatic-loudspeakers.html

 

[PaulD]

Lovely work! Also thanks for all the info and links!

 

[tomtoo]

WOW! Thats what i would call a advanced DIY project.

 

[Jazzman53]

Thanks you guys for the compliments on my speakers. I can mostly claim credit for the woodwork but this project would not have been possible without a lot of help from ESL gurus 'Bolserst', 'Golfnut' and 'Calvin' at the DIY Audio Forum. And I'm just happy to pay it forward.

BTW; my website below is an extensive compilation of photos and information on building ESL speakers:
http://jazzman-esl-page.blogspot.com/

 

[Hayabusa]

Did you segment the speaker to have less beam forming for the high frequencies?

 

[Jazzman53]

Did you segment the speaker to have less beam forming for the high frequencies?

Yes.

I love the slam and imaging of an unsegmented panel but its ultra-narrow, tightly focused sweet spot (sometimes referred to as the "head-in-a-vise" effect) is rather inconvenient when company drops in for a listen, so I opted for wider dispersion using a segmented panel. And using many small segments as opposed to a few wider ones, gives very smooth trending dispersion.

Bending the wavefront can be done passively by merely inserting the appropriate value resistors between the wire groups. These resistances in series couple with the wire groups' capacitances in parallel to form a series of low pass filters which progressively chop off the upper frequencies toward the panel edges. And the charging-time of the wire groups' capacitances progressively attenuates phasing towards the panel edges as well. The combined effect bends the wavefront from planar to curved.

A very smart ESL guru (and PHD physicist); Rod White (a.k.a. 'Golfnut' on the DIY Audio Forum) published a technical paper on how to do this. And his collaborator; Steve Bolser, has created an Excel spreadsheet program which does all the math. I just plug in my panel dimensions, diaphragm-to-stator gap, biasing voltage, and transformer step-up ratio and the spreadsheet automatically calculates the resistor values, SPL, and generates a frequency response graph.

The link/image below depicts dispersion pattern simulations from a flat panel vs a 30-degree arc curved panel versus a finely segmented flat panel: Steve+B.jpeg

 

[Eurasian]

What a craftsman! Excellent dispersion as well! Any distortion data available?

 

[Hayabusa]

Yes.

The link/image below depicts dispersion pattern simulations from a flat panel vs a 30-degree arc curved panel versus a finely segmented flat panel: Steve+B.jpeg

So this method gives exactly the same results as a 30 degree curved panel?
Nice! could this angle even be extended by an even wider panel with more segments?

 

[Jazzman53]

So this method gives exactly the same results as a 30 degree curved panel?
Nice! could this angle even be extended by an even wider panel with more segments?

No... better. A properly segmented panel gives wider and much smoother trending dispersion as compared to a 30-degree arc panel of similar height/width.

 

[Jazzman53]

What a craftsman! Excellent dispersion as well! Any distortion data available?

Sorry; I'm not setup to do that measurement. I have quite a lot of tuning flexibility with my Behringer DCX2496 crossover and DEQ2496 EQ but no means to measure distortion.

ESLs typically have almost no distortion other than the diaphragm's fundamental [drum-head] resonance. A big advantage for a hybrid ESL is that the panel is typically crossed over at least an octave above the drum head resonance; so the panel isn't being actively driven at that frequency. Still; any woofer operating anywhere in the room will excite the diaphragm's drum head resonance to some extent. A distortion free speaker doesn't exist but ESLs can do better than most.

When I first heard about electrical segmentation; I was skeptical that inserting high value resistors between the wire groups wouldn't add audible distortion. But after several years of listening to both segmented and unsegmented panels; aside from wider dispersion and less EQ'ing required, I cannot discern any audible differences.

A friend described my speaker's sound as "spooky".

 

[Jazzman53]

Forgot to mention:
I've done A/B comparisons of the sound of segmented versus unsegmented panels using the very same panel operating in switch-selectable narrow and wide dispersion modes. This panel used welding rod conductors with a multi-pole rotary switch inserted to enable jump-over the segmentation resistors (for narrow mode) or leave them engaged (for wide mode).

The switch mode feature was fun but the novelty wore off fairly quickly because the switch was on the HV side so I had to power down to switch modes (to avoid frying the switch) and then re-EQ the panel because changing modes also changed the response curve.

Here's a link to that project: http://jazzman-esl-page.blogspot.com/2008/03/a-segmented-wire-stator-esl-with.html

 

[Jazzman53]

More porn for the planar pervs:

 

[PaulD]

Phil Spector's Wall of Sound!

Awesome work there Jazzman!

 

[Doodski]

Right on! That's home building on steroids. Nice work!

 

[Blumlein 88]

More porn for the planar pervs:
View attachment 54646

Ok, so what is the design for upper speakers in your ESL Atmos setup?

 

[PaulD]

A fully addressable panel array across the whole ceiling!