DIY Electrostatic Speakers

[Jazzman53]

Hi all,
I built my 'beam splitter' speaker eleven years ago, and it's electronics and stat panels have evolved quite a bit in the years since. Originally it had perf-metal stators, then more efficient micro-perf stators, then segmented welding rod stators, and now state of the art segmented wire stators that electrically curve the wave front. The links below explain it all, with build photos, and If anyone is interested; I am happy to share my CAD drawings. Enjoy!

Jazzman's DIY Electrostatic Loudspeaker Page
Video of the new "Audi" speaker at Carverfest 2017

 

[andreasmaaan]

Hey, I've come across your design online before. Very impressive, and thanks for sharing

I'm curious about the use of highly reflective wood (the housing for the transmission line) apparently just a few centimetres behind the electrostatic panel. I would have thought the effect of this would be to reflect forward the backwave from the panel, which would then recombine with the front wave (with a short delay). Do you consider this a feature or a drawback, and if the latter, how do you try to deal with it in the design?

I also wonder if you've published any measurements of the current design anywhere?

 

[Jazzman53]

I had the concern that you described before committing to the build. I did some listening tests with and without the beam splitter behind the panel and I couldn't hear anything objectionable; only a change in the room ambience; which also was not objectionable. I suspect the beam splitter adds a bit of impedance, but the bounce back is apparently being deflected to the side rather than back to the diaphragm. I was amazed that it didn't seem to have an adverse affect. And the speaker does sound pretty incredible.

The beam splitter turned out to be an elegant solution to the problem of where to stuff all that box volume, and for years afterward I had thought it was my original idea-- until I saw a drawing of an ESL built 30 years ago by the Shackman company in Germany. The Shackman speaker was smaller with two woofers, one on top and one below a V-section beam splitter, and the ESL panel in front of the splitter was curved!!!! Assuming the back wave's direction of travel is normal to the diaphragm, a curved diaphragm would be much closer to parallel with the beam splitter's surfaces, and I figure that had to result in some bounce back and standing waves between them. Whereas, a flat panel would not be parallel to a V-section beam splitter behind it.

I would love to post some measurements but I don't have the measurement gear or the know how to use it. It's an old dog, new trick thing.... I read up on the Room EQ software but it just made my brain hurt.

 

[Jazzman53]

It's about time I updated this post.

My original 'Beam Splitter' ESL shown above found a new home with my friend Eric in Kentucky.

I have since built a third generation ESL design, which I call the Jazzman MkIII. The MkIII is fully dipolar, with the woofer mounted on an open baffle below the stat panel. It's also electrically segmented to give wider, smoother trending dispersion than a curved panel. I've also added a new pair of RiPol subs for the bottom end.

Full details and build pics are shown on my website here: Jazzman's DIY Electrostatic Loudspeaker Page

This is my masterpiece speaker, and I'm now retired from speaker building:

Below: With magnetically attached grill covers on:

 

[JohnnyNG]

I've read your blog and am amazed at your project and talent - so impressive. "Masterpiece" is not hyperbole.

I could never plan and tackle something like that but I do enthusiastially agree on the Eminence Kappalite 3012LF. I'm running a pair for mid/upper bass in sealed cabs actively crossed between 80-300Hz and they have been absolutely transformative. I've never heard/felt such lifelike and visceral punch and slam outside a live venue.

3 way active with real punch.. which way would you go.? Pro driver or .?

I'm very impressed by Audiohorn's approach to horn design, though I'm still plowing through this. https://audiohorn.net/sciences/critical-distance/ Meanwhile, what are the criteria for choosing between the X-40 horn and the biradial horn? Audiohorn recommends specific drivers to use with...

www.audiosciencereview.com

Again, incredible work!

 

[OldHvyMec]

I like everything except the sub, I'll pass on that one. The panel looks, and I'm sure with the back side blocked, sounds wonderful. That has been my experience over the last 40 years with bi-directional drivers, and there are no exceptions for mids/highs or 250-300 Hz and >. Subs I've always enjoyed the OB subs that Rythmik servo plates, GR drivers, and DIY cabinets bring to the table. I also like Subs that use PRs for fine-tuning the sub to the room, amp, and cables.

I've never seen a system that uses a subs not benefit from the DBA concept, though I have enjoyed a single well-placed sub in many systems. The only issue is that a single sub usually offers ONE good seated listening position. I personally have zero problem with that. I don't need great bass throughout the entire room, though I enjoy great sub/bass in a room built for dancing.

Very nice cabinet work. 10+ on that

My best regards.

 

[Jazzman53]

I've read your blog and am amazed at your project and talent - so impressive. "Masterpiece" is not hyperbole.

I could never plan and tackle something like that but I do enthusiastially agree on the Eminence Kappalite 3012LF. I'm running a pair for mid/upper bass in sealed cabs actively crossed between 80-300Hz and they have been absolutely transformative. I've never heard/felt such lifelike and visceral punch and slam outside a live venue.

3 way active with real punch.. which way would you go.? Pro driver or .?

I'm very impressed by Audiohorn's approach to horn design, though I'm still plowing through this. https://audiohorn.net/sciences/critical-distance/ Meanwhile, what are the criteria for choosing between the X-40 horn and the biradial horn? Audiohorn recommends specific drivers to use with...

www.audiosciencereview.com

Again, incredible work!

I have to agree: The Eminence KappaLite 3012 woofer truly is transformative! It's punch is phenomenal!

Although the KappaLits's TS parameters are optimized for a ported box, I'm using it on an open baffle (actually a sort of modified H-baffle) to match the ESL's dipole radiation pattern, and its low-Q, low-inductance, and tight suspension allow it to blend seamlessly with the near-zero-Q electrostat.

The woofer/panel integration is the best I've heard in a hybrid ESL.

 

[JohnnyNG]

Although the KappaLits's TS parameters are optimized for a ported box, I'm using it on an open baffle (actually a sort of modified H-baffle) to match the ESL's dipole radiation pattern, and its low-Q, low-inductance, and tight suspension allow it to blend seamlessly with the near-zero-Q electrostat.

I can only imagine! Yeah, the 3012LF won't go very low on its own sealed (although Eminence does suggest a couple sealed designs) but, like you, I only needed it to go down far enough to cross to the dedicated subwoofer.

 

[Zinda]

Hi all,
I built my 'beam splitter' speaker eleven years ago, and it's electronics and stat panels have evolved quite a bit in the years since. Originally it had perf-metal stators, then more efficient micro-perf stators, then segmented welding rod stators, and now state of the art segmented wire stators that electrically curve the wave front. The links below explain it all, with build photos, and If anyone is interested; I am happy to share my CAD drawings. Enjoy!

Jazzman's DIY Electrostatic Loudspeaker Page
Video of the new "Audi" speaker at Carverfest 2017

View attachment 26274

When I was testing my DIY design using JansZen panels I came up with a very similar idea that would eliminate the rear wall from the equation by adding a curved back plate behind the cluster of panels. The idea was to reflect off the back panel and bring the sound out the sides top and bottom which would be aimed directly at your head. I never got around.d to doing any real tests before they melted in the fire. I'll have to transfer some pics to my phone. Is that 19 segments or 87 segments or divisions?

I'm currently looking into refiguring the woofers in the Prodigy. Big dip at 250Hz (supposed crossover point but I haven't tested it yet) even when I reversed the woofers input. So that means it's a problem within their strange bandpass circuit design. 2 different scanSpeak 10" drivers that play different frequencies. I'll bypass their crossover and test both running full range wired in phase and out to find out what's going on. I'm testing the panels lowest frequency again, last time it looked like it was 250 at roll off start.

Bass just sounds wrong. Everything else is good, I do need to hunt down a can vibration, even my homemade Heresy cans are noise free, you'd think at $10k they'd be able to design a noise free cab and have huge bass reaching I to the teens!

 

[Heinrich]

I have to agree: The Eminence KappaLite 3012 woofer truly is transformative!

Lately I raised a question regarding possible mirror charges on the stators. Additionally I was curious how the membrane would break up (which it necessarily does). Both effects would lead to chaotic behavior of the membrane, and so the question arises, how such would be controlled. It cannot use the air load, as that would necessarily lead to sound emission, counteracting the intent to disguise the irregularities.

In short, are there any measurements on frequency response first, and second on the relative phase of different parts of the diaphragm? I assume that would be of top-notch interest. Thanks anyway for sharing!

 

[Jazzman53]

Lately I raised a question regarding possible mirror charges on the stators. Additionally I was curious how the membrane would break up (which it necessarily does). Both effects would lead to chaotic behavior of the membrane, and so the question arises, how such would be controlled. It cannot use the air load, as that would necessarily lead to sound emission, counteracting the intent to disguise the irregularities.

In short, are there any measurements on frequency response first, and second on the relative phase of different parts of the diaphragm? I assume that would be of top-notch interest. Thanks anyway for sharing!

A typical tall, slender, unsegmented flat panel ESL will exhibit a rising frequency response due to the combined effects of beaming highs and the dipole phase cancellation rolling off the lows.

My panel is atypical in that it employe symmetrical segmentation (15 discrete wire groups arrayed in 7+1+7 symmetry) for wide dispersion. The segmentation scheme also effectively EQ's the panel's frequency response all the way down to the LF cutoff.

I don't have REW or other means to plot a graph my speaker's frequency response. However, the attached screenshot shows my DSP's RTA reading pink noise from the speakers, picked up at my listening position. The panel response starts rolling off at around 15kHz, and I could compensate that with a parametric EQ but I can't hear that high anyway so I don't worry about it.

Other than measuring the diaphragm's drum-head resonance, which occurs at about 100Hz in my speaker, I have not attempted to measure or quantify the diaphragm's vibrational modes. However, I do believe the vibrational modes are significantly dampened by the air load, as I believe the coupled air mass is significantly greater than that of the 6-micron diaphragm.

In fact; I belive an ESL's characteristic clarity is predominantly due to the diaphragm being so effectively dampened by the coupled air mass (i.e. it just doesn't ring to the extent that a more massive moving element would).

 

[Heinrich]

... the diaphragm being so effectively dampened by the coupled air mass (i.e. it just doesn't ring to the extent that a more massive moving element would).

Thanks! It's a bold move to design and construct a panel w/o having the opportunity to measure the outcome. Some would scrutinize even their el cheapo wideband single driver mini boxes to death ;-) Bold and successful, nice!

 

[JohnnyNG]

Some would scrutinize even their el cheapo wideband single driver mini boxes to death ;-)

That's the part of the hobby many seem to enjoy the most, and I can appreciate that, but I can also understand just wanting to sit back and listen once the basics are checked.

 

[Jazzman53]

Since my first ESL build in 2008, which was an unsegmented flat panel design, I've built about 20 additonal pairs for myself and others, using the same panel dimensions but different segmentation schemes and woofer alignments.

I didn't feel I was taking a big gamble with the MkIII panel design, which I consider a refinement of an already proven configuration. The biggest change from previous versions was the OB woofer, which was carried over from a speaker I designed and built for my friend Martin.

I named Martin's speaker the "Audi" design because he required that the 5'6" tall speakers be transportable in the back of his 2010 Audi sports car (rear seat folded forward and speakers inserted through the trunk). Boxed bass alignments were too big to fit in the car but an OB alignment fit with clearance.

Woofer/panel integration (the bane of most hybrid ESLs) was so seamless in Martin's speakers that I used the same OB and woofer in the MkIII design.

Of course, once I had built speakers for one friend, I could not decline requests from my other audio friends who had heard Martin's pair at the CarverFest retreat. So I ended up building six additional pairs.

The last four pairs were built concurrently, and consumed my evenings and weekends for a full year. The four-pair build (see below) was so exhausting, it completely healed my speaker building psychosis and I've sworn to build no more.

Porn for planar pervs:

 

[Zinda]

It's about time I updated this post.

My original 'Beam Splitter' ESL shown above found a new home with my friend Eric in Kentucky.

I have since built a third generation ESL design, which I call the Jazzman MkIII. The MkIII is fully dipolar, with the woofer mounted on an open baffle below the stat panel. It's also electrically segmented to give wider, smoother trending dispersion than a curved panel. I've also added a new pair of RiPol subs for the bottom end.

Full details and build pics are shown on my website here: Jazzman's DIY Electrostatic Loudspeaker Page

This is my masterpiece speaker, and I'm now retired from speaker building:

View attachment 460720

Below: With magnetically attached grill covers on:
View attachment 460721

I love the simplicity, the open back woofer seems like a common sense decision that would work in the same way as the panel. I have always felt that if there is a lack of low end with a single boxed driver adding another facing either direction but out of phase would help 2 fold, first reinforcing the panels rear waves to include more full range response and if the woofers are mounted in an isobaric configuration, there could be a gain in lowest frequency output.

Of course things don't always work in real life and need real world testing to prove any benefit, the problem comes in when testing tools cannot determine overall timber and subjective blending that affects imaging and soundstage which are where ESL panels shine. I don't think REW can do much to determine hybrid design other than to check overall flat response and as I have found, voicing the panels to get an exact equal response from both channels. I find this to be a very critical step. Placement will change what frequencies land at the listening position and how the imaging and soundstage will shift at certain frequencies.

I used to have a dual 31 band EQ within arms reach of the listening position and could dial it in over time. This could take weeks before a perfect response is achieved. Of course there's more to it than random adding at certain frequencies since adjusting will change the overall sound plot from your original flat output. This requires doing more overall checks as you proceed. That's where the extra time comes in.

I have to add that flat plots on each channel don't mean that the overall combined plot will remain flat and simply show a gain in SPL. Sometimes it gets tricky to get a perfect balance between each channel and the combined output. With waves cancelling at the listening position, that's where speaker position plays the biggest the biggest role and will force you to start the process over..

I may be expecting more out of my system than others, but I have realized exactly what's possible from a sound system by doing this, I have discovered limitations in control using 1 idea or another to get the exacting results expected. Positioning is the hardest part, moving a speaker just an inch will change listening position perfection quite a bit. I found ESL's have a more forgiving overall plot than box speakers but have a fragile imaging and soundstage where they excel. Finding the perfect placement and dialing in voicing can become a never ending task since each affects the other in unexpected ways.

There's been more than once where I have given up and removed the EQ and I have also used passive crossovers and found the sound to be good enough, but it never lasts long.