Do you owe it to yourself to check out electrostatic speakers?

[Jazzman53]

Do you owe it to yourself to check out the latest electrostatic speaker designs?

Ahh; YES.

No speaker is perfect but I wish everyone could hear the latest commercial and DIY ESL designs.
I'm sure many would find their sound addictive.

I will discuss both, but first I must confess my bias favoring ESLs, which I've championed for years, and in particular my own DIY design.

In my experience, it's far easier for a DIY'er to build a world-class ESL than a world-class conventional speaker. I don't mean that it's less work (the labor is horrendous)-- just that a good result is more certain because you start with less baggage.

I can list as many cons as pros for ESLs but the pros are really good and the cons are not terribly hard to remedy in a good design. Some manufacturers mitigate the cons quite well and the latest offerings are better than ever.

Pros:
1. The ultra-low moving mass diaphragm is so effectively dampened by the greater air mass it's coupled to, that it simply doesn't *ring.
2. Line-source configuration gives really good imaging.
3. Contrary to popular belief and the commercial offerings, DIY designs can tailor dispersion patterns as desired or even provide switch-selectable wide & narrow dispersion modes (costly but do-able).
4. No crossover-- or a hybrid can use a single crossover down low where it's less audible.
5. Typically excellent/superior coherence, resolution & transient response.

*An ESL diaphragm is remarkably resonance-free across the audio bandwidth except for its single fundamental "drum-head" resonance peak which typically occurs in the bass region at a frequency determined by its tension. This resonance is analogous to a woofer's free-air resonance (f/s) and is likewise undesirable.

Cons:
1. The diaphragm's drum-head resonance is a nasty-loud distortion that's difficult to mask.
2. Anemic bass unless the panel is huge, and drum-head resonance can render it boomy & sluggish.
3. Hybrids add the bass but few achieve a seamless blending of the woofer & stat panel.
4. Narrow dispersion which some listeners find objectionable (i.e. "head-in-a-vise" effect).
5. Capacitive load & impedance as low 1 Ohm in treble band requires a strong, stable amp.
6. Long-term reliability is historically iffy, especially in DIY versions.

Mitigating the drum head resonance:
Acoustat and some other full range designs apply a felt strip or mesh fabric on the rear stator which partially absorbs the resonance energy.

Soundlab uses the drum-head resonance to boost bass output by dividing the diaphragm into multiple different-width sections which break up the single/loud/narrow resonance peak into multiple softer resonances spread over a wider bandwidth (i.e. "distributed resonance"). It's an innovative solution that mitigates the booming effect and boosts bass output, but it's not a perfect solution because the the drum head resonance is by definition a distortion, so using it must include its distortion.

Martin Logan tensions the diaphragm quite high, and likewise uses distributed resonance to enhance the upper bass/lower midrange, so that less EQ'ing is required to offset the panel's dipole roll off and flatten its response curve. Here again, distributed resonance spreads out the drum-head resonance, which necessarily includes its distortion.

Mitigating the narrow dispersion:
Narrow dispersion is not necessarily a flaw. In fact, the Sanders/InnerSound speakers are purposely designed or narrow dispersion because it gives the most precise imaging at their tightly focused sweet spot. Some prefer the narrow sweet spot but it's not so good for entertaining guests, and most speaker designers opt for wider dispersion.

Some designs (ML, Soundlab) curve the stat panel to about 30 degrees of arc, which curves its projected wave-front and widens the sweet spot. The small curvature is still a bit beamy but retains much of the magical imaging.

Some designs use a separate, narrow treble panel to spread the highs.

The Quad 63 and similar designs use separate bass panels and a center mid/treble panel with concentric ring conductors sequentially driven thru an LC delay line to function as a point source projecting a spherical wave front.

Mitigating the load capcitance:
Most designs add low value resistors in front of the transformers which renders the load partially-resistive (although still mostly capacitive).

Reliability:
Most manufactures now produce reliable speakers, although 20 years seems to be the best-case limit (except for Acoustats which are seemingly immortal). I would not advise purchasing a DIY ESL unless you do your homework on the specific design, and you and trust the builder to stand behind it.

The weakest link in any ESL is the diaphragm coating degrading over time. So; a good design goal would make the panel easy to disassemble without damaging the stators, and preferably also leaving the diaphragm intact. If thus accessible; the diaphragm could be easily re-coated or replaced.

The most robust designs use insulated wire stators because of their higher resistance to arcing.
Perf metal stators can be reliable if made by a manufacturer like ML who has mastered the specialized processes required to prevent arcing.

I would no longer build or purchase a DIY ESL with perf metal stators because it's so difficult to perfectly smooth-over the cut metal edges and apply the insulating coatings as required to prevent arcing.

The old Acoustat ESLs are practically bullet proof because they used superbly insulated wire stators and a simple carbon black diaphragm coating that remained stable over time-- these continue to play happily for decades!

So; if you want a reliable ESL; buy or build one similar to the Acoustat design. Or at least with stators than can be easily disassembled to access the diaphragms.

*********************************************************************************************************************

My latest DIY design (Jazzman MkIII):
The drum head resonance: Is simply avoided (not excited) by tensioning the diaphragm to resonate <100Hz, and then crossing the panel out to the woofer at least one octave above the resonance using a steep-sloped digital crossover (250Hz @48db/oct). Hence; the panel is not being energized/excited near the diaphragm's resonance frequency.

Dispersion: Tailoring the dispersion requires choosing either a narrow sweet spot with very precise imaging or a wide sweet spot with less precise imaging or something in-between (physics forces this choice-- you can't have it all).
I chose the latter option and tailored the segmentation scheme accordingly.

The stat panel employs symmetrical multi-segmented wire stators driven sequentially from the panel centerline outward to function as a line source projecting a cylindrical wave-front. This works amazingly well; giving wide, smooth-trending dispersion.

Load/Impedance: The RC line used to curve the wave front effectively linearizes the impedance and renders it predominantly resistive, as only the smaller capacitance of the first two wire groups is reflected back to the amp. This also reduces the degrees of phase rotation for the impedance reflecting back to the amp (very desirable).

Woofer/Panel Integration:
The matching dipole radiation patterns of the excellent OB mounted Eminence KappaLite 3012LF woofer and the stat panel merge seamlessly to sound like a single, coherent driver-- the best integration I've heard in a hybrid ESL.

Reliability:
The wood lattice supported wire stators are less impact-tolerant than perf metal stators and must be handled accordingly. Otherwise; they are highly arc resistant resistant and therefore unlikely to be harmed by even crazy-loud volume levels. Even with no protection circuitry, the electronics interface has also proven to be reliable. Over the past five years I've built seven pairs for myself and audio pals, and none have failed.

The stators can be either bond bonded together or mechanically mated in this design. If mechanically mated; the stators can be easily separated to access the diaphragms, should servicing them ever be needed.

How long will the diaphragm coating last? I can vouch for 5 years so far-- but I'll get back to you on that in about 10 years if I'm blessed to still be here.

Website with build photos & write-ups: http://jazzman-esl-page.blogspot.com/?m=0

Photo of the Jazzman MkIII ESL:

 

[Keith_W]

I spent a bit of time reading your blog. Fascinating journey, I did not know that ESL's could be DIY. Couple of questions for you:

- What are your crossover points?
- Do you have any measurements? Also, if you did take measurements, where did you place the microphone?
- I have heard that Quad ESL's are able to mimic a point source because they arrange their stators in a concentric ring. What are your thoughts on this?
- Those cardioid subs are interesting, is there a reason you chose them for your system instead of more conventional subs?

 

[Jazzman53]

Hi Keith,

Thanks for your reply.

I attempted to learn the REW software but I've grown too old and senile to learn such things quickly and the setup would have required changing my system back to analog signal path with an analog preamp in the loop, which I didn't want to bother with so... no, I unfortunately have no measurement data or graphs to share.

I do however tune the system with a mic and the auto-tune feature in my DBX Driverack Venu 360 crossover.

I first setup symmetric-sloped crossovers at 70Hz & 260Hz using the 48db/octave Linkwitz-Riley filters.

I then time align the woofer and stat panel using the mic placed about 2ft way and between the woofer and panel. I feed in a 260Hz test tone (matching the crossover frequency) and adjust the panel delay to obtain max SPL on the RTA (i.e. max constructive interference).

Next I balance the channel gains using the RTA with pink noise.

I then run the auto-tuning wizard and mic sweeps, and the DBX runs the calculations and overlays parametric EQ's to match the response to my pre-selected curve. The I-pad app's graphic display shows where to place the mic for each sweep, and there's an option to select either 3 or 4 sweeps for the auto-tuning algorithm. I can't say enough good things about the Venu 360-- I love this thing!

I do the final tuning by ear with music: First I play the Erin Bode tune "Holiday" and use the graphic EQ to take the edge off her voice (about 3db down at 2.5kHz). Erin has a clear high voice that cuts diamonds if the tuning is hot, so she's my go-to track for that purpose. I then follow up with Brubeck's "Take Five" and if Morello's hi-hat is still crisp, then I know I haven't taken too much off with Erin's voice track. That pretty much gets me happy.

You mentioned the Quad 63 and its point source projection. I heard a pair of those about 15 years ago and it seemed to function as advertised and sounded really good... a great improvement over the much revered Quad 57 in my opinion.

Likewise; the segmentation scheme in my ESL works amazingly well. If you put your ear close to the panel you can clearly discern the treble radiating exclusively from the panel centerline, and with 15 wire groups in 7+1+7 symmetry, the dispersion pattern is smooth-trending (more wire groups = smoother trending dispersion).

The Ripol subs have an a unique sound that I absolutely love. It took me a while to overlook their inefficiency and fully appreciate what they do and how they do it. They produce simply the cleanest, tightest bass I've ever heard-- but they are definitely quality over quantity.

Before I built the Ripols I used a single monster-powerful long-throw Dayton 12" sub in a large (2.8ft3) cab. This thing would bust up dry wall if cranked it up, but it always sounded boomy and I could never get it to blend seamlessly with the mains-- not even close. I finally realized that the booming was predominantly the room resonance but I could never position the sub to tame it.

Replacing the box sub with the Ripols was like magic-- the booming disappeared and the tone was deep and lovely. I have them crossing in down low (70Hz) with a steep filter slope so I seldom notice them playing until I put on a tune with profound bass.

Most music doesn't have a lot of low bass content so, with the Ripols crossing in so low and being so clean and unobtrusive, I seldom noticed them on most tracks. Even so; if I kill the sub amp the foundation of the music disappears and I go "oh yeah" they were there and doing their thing all along.

I play something with profound bass like Bela Fleck's "Flight of the Cosmic Hippo" or Malia & Boris' "Convergence" album, the Ripols come alive and I go "Wow! that's Deep...and Low!" Not chest thumping but clean and tight like I've never experienced. The bass messiah has come down from heaven

I don't know that I would recommend Ripols for loud rock but they are wonderful for Jazz!

Charlie

 

[RayDunzl]

How long will the diaphragm coating last? I can vouch for 5 years so far-- but I'll get back to you on that in about 10 years if I'm blessed to still be here.

My ML reQuests are 25 years old now.

I've kept the panels covered with a well-fitting jersey body pillow case.

 

[gnarly]

Hi, thx for your post. Good read !

I've been an electrostat fan for over almost 50 years. I've always tried to listen to any i could manage to audition.
They simply have a uniquely beautiful sound to my ears.

Still have a pair of old Acoustat-X running, with the servo amps.
Jim Strickland refurbished the amps not too long ago. THEY STILL ROCK and with finer finesse than is seldom heard in today's speakers...
.......and maybe/hopefully are nearly immortal as you say!

Super impressed with your DIY work.....go Charlie go !!!!!!!

 

[dlaloum]

With years of owning Quad ESL's (first the 57's, then the 63's and the 989's) - yes there are things that ESL's do that "normal" speakers really struggle with - and if those are important to you (and their weaknesses less so) then ESL's are a great way to high end audio.

WAF and space constraints combined to cause me to move away from my Quads (I still have a pair of ESL57's in storage).

In searching for an acceptable alternative I found that the Gallo speakers with their CDT tweaters, and mechanically tuned crossover (ie: no crossover, but speakers designed to cross over to each other through their natural properties) - achieved 99% of what my ESL63/989's could do ....

If I was looking today, I would be investigating the various new tweeter types and seeking out similar crossoverless designs (unless I had household approval for ESL's.... in which case I would probably go back to Quads!!)

 

[JohnnyNG]

The speakers and blog are awe-inspiring. Wow.

 

[ferrellms]

Do you owe it to yourself to check out the latest electrostatic speaker designs?

Ahh; YES.

No speaker is perfect but I wish everyone could hear the latest commercial and DIY ESL designs.
I'm sure many would find their sound addictive.

I will discuss both, but first I must confess my bias favoring ESLs, which I've championed for years, and in particular my own DIY design.

In my experience, it's far easier for a DIY'er to build a world-class ESL than a world-class conventional speaker. I don't mean that it's less work (the labor is horrendous)-- just that a good result is more certain because you start with less baggage.

I can list as many cons as pros for ESLs but the pros are really good and the cons are not terribly hard to remedy in a good design. Some manufacturers mitigate the cons quite well and the latest offerings are better than ever.

Pros:
1. The ultra-low moving mass diaphragm is so effectively dampened by the greater air mass it's coupled to, that it simply doesn't *ring.
2. Line-source configuration gives really good imaging.
3. Contrary to popular belief and the commercial offerings, DIY designs can tailor dispersion patterns as desired or even provide switch-selectable wide & narrow dispersion modes (costly but do-able).
4. No crossover-- or a hybrid can use a single crossover down low where it's less audible.
5. Typically excellent/superior coherence, resolution & transient response.

*An ESL diaphragm is remarkably resonance-free across the audio bandwidth except for its single fundamental "drum-head" resonance peak which typically occurs in the bass region at a frequency determined by its tension. This resonance is analogous to a woofer's free-air resonance (f/s) and is likewise undesirable.

Cons:
1. The diaphragm's drum-head resonance is a nasty-loud distortion that's difficult to mask.
2. Anemic bass unless the panel is huge, and drum-head resonance can render it boomy & sluggish.
3. Hybrids add the bass but few achieve a seamless blending of the woofer & stat panel.
4. Narrow dispersion which some listeners find objectionable (i.e. "head-in-a-vise" effect).
5. Capacitive load & impedance as low 1 Ohm in treble band requires a strong, stable amp.
6. Long-term reliability is historically iffy, especially in DIY versions.

Mitigating the drum head resonance:
Acoustat and some other full range designs apply a felt strip or mesh fabric on the rear stator which partially absorbs the resonance energy.

Soundlab uses the drum-head resonance to boost bass output by dividing the diaphragm into multiple different-width sections which break up the single/loud/narrow resonance peak into multiple softer resonances spread over a wider bandwidth (i.e. "distributed resonance"). It's an innovative solution that mitigates the booming effect and boosts bass output, but it's not a perfect solution because the the drum head resonance is by definition a distortion, so using it must include its distortion.

Martin Logan tensions the diaphragm quite high, and likewise uses distributed resonance to enhance the upper bass/lower midrange, so that less EQ'ing is required to offset the panel's dipole roll off and flatten its response curve. Here again, distributed resonance spreads out the drum-head resonance, which necessarily includes its distortion.

Mitigating the narrow dispersion:
Narrow dispersion is not necessarily a flaw. In fact, the Sanders/InnerSound speakers are purposely designed or narrow dispersion because it gives the most precise imaging at their tightly focused sweet spot. Some prefer the narrow sweet spot but it's not so good for entertaining guests, and most speaker designers opt for wider dispersion.

Some designs (ML, Soundlab) curve the stat panel to about 30 degrees of arc, which curves its projected wave-front and widens the sweet spot. The small curvature is still a bit beamy but retains much of the magical imaging.

Some designs use a separate, narrow treble panel to spread the highs.

The Quad 63 and similar designs use separate bass panels and a center mid/treble panel with concentric ring conductors sequentially driven thru an LC delay line to function as a point source projecting a spherical wave front.

Mitigating the load capcitance:
Most designs add low value resistors in front of the transformers which renders the load partially-resistive (although still mostly capacitive).

Reliability:
Most manufactures now produce reliable speakers, although 20 years seems to be the best-case limit (except for Acoustats which are seemingly immortal). I would not advise purchasing a DIY ESL unless you do your homework on the specific design, and you and trust the builder to stand behind it.

The weakest link in any ESL is the diaphragm coating degrading over time. So; a good design goal would make the panel easy to disassemble without damaging the stators, and preferably also leaving the diaphragm intact. If thus accessible; the diaphragm could be easily re-coated or replaced.

The most robust designs use insulated wire stators because of their higher resistance to arcing.
Perf metal stators can be reliable if made by a manufacturer like ML who has mastered the specialized processes required to prevent arcing.

I would no longer build or purchase a DIY ESL with perf metal stators because it's so difficult to perfectly smooth-over the cut metal edges and apply the insulating coatings as required to prevent arcing.

The old Acoustat ESLs are practically bullet proof because they used superbly insulated wire stators and a simple carbon black diaphragm coating that remained stable over time-- these continue to play happily for decades!

So; if you want a reliable ESL; buy or build one similar to the Acoustat design. Or at least with stators than can be easily disassembled to access the diaphragms.

*********************************************************************************************************************

My latest DIY design (Jazzman MkIII):
The drum head resonance: Is simply avoided (not excited) by tensioning the diaphragm to resonate <100Hz, and then crossing the panel out to the woofer at least one octave above the resonance using a steep-sloped digital crossover (250Hz @48db/oct). Hence; the panel is not being energized/excited near the diaphragm's resonance frequency.

Dispersion: Tailoring the dispersion requires choosing either a narrow sweet spot with very precise imaging or a wide sweet spot with less precise imaging or something in-between (physics forces this choice-- you can't have it all).
I chose the latter option and tailored the segmentation scheme accordingly.

The stat panel employs symmetrical multi-segmented wire stators driven sequentially from the panel centerline outward to function as a line source projecting a cylindrical wave-front. This works amazingly well; giving wide, smooth-trending dispersion.

Load/Impedance: The RC line used to curve the wave front effectively linearizes the impedance and renders it predominantly resistive, as only the smaller capacitance of the first two wire groups is reflected back to the amp. This also reduces the degrees of phase rotation for the impedance reflecting back to the amp (very desirable).

Woofer/Panel Integration:
The matching dipole radiation patterns of the excellent OB mounted Eminence KappaLite 3012LF woofer and the stat panel merge seamlessly to sound like a single, coherent driver-- the best integration I've heard in a hybrid ESL.

Reliability:
The wood lattice supported wire stators are less impact-tolerant than perf metal stators and must be handled accordingly. Otherwise; they are highly arc resistant resistant and therefore unlikely to be harmed by even crazy-loud volume levels. Even with no protection circuitry, the electronics interface has also proven to be reliable. Over the past five years I've built seven pairs for myself and audio pals, and none have failed.

The stators can be either bond bonded together or mechanically mated in this design. If mechanically mated; the stators can be easily separated to access the diaphragms, should servicing them ever be needed.

How long will the diaphragm coating last? I can vouch for 5 years so far-- but I'll get back to you on that in about 10 years if I'm blessed to still be here.

Website with build photos & write-ups: http://jazzman-esl-page.blogspot.com/?m=0

Photo of the Jazzman MkIII ESL:
View attachment 294103

Cool looking and congrats on implementing your own design!

Having had Acoustats (20 years ago) I cannot agree that electrostatics are accurate for listening (beamy and not the most accurate frequency response), although they can sound fantastic with just the right listening material, room placement, and if your head is just in the right place.

 

[Jazzman53]

My ML reQuests are 25 years old now.

I've kept the panels covered with a well-fitting jersey body pillow case.

If you search the ML owners forum you will find more than a few posts lamenting panels less than 10 years old losing volume. But you will also find responding posts reporting much older panels that still play fine. So I'm wondering if environmental factors are affecting the diaphragm coating. ML's don't use dust covers so UV light exposure could be a factor--- maybe you should keep your pillow cases on!

My speakers have magnetically attached grills on front and back. I show them in posts with the grills off to display their wire stators but the grills are acoustically transparent so I usually leave them on, even when playing music:

 

[anmpr1]

Still have a pair of old Acoustat-X running, with the servo amps. Jim Strickland refurbished the amps not too long ago. THEY STILL ROCK and with finer finesse than is seldom heard in today's speakers...

I am happy to learn that Jim is still involved. I owned several Acoustat loudspeakers, his TNT 200 amplifier, plus a TNP preamp.

For what they offered, I considered them good value. Of course with his loudspeakers you had the 'head in a vice' image situation, but that's how it is if you choose ESL.

The only exception to that rule that I ever encountered was Harold Beveridge's creation, with its unique 180 degree acoustic lens. But they were out of my price range. Their unusual (albeit appealing) sonic presentation was something I'd never quite encountered. It was like you were somehow 'immersed' in the sound. Kind of weird, actually. However that was, I don't know if it was really what I was after; or whether it was long-term viable. Additionally, its integrated electronics were rather tweaky.

Acoustat, probably more than any other company (except possibly early Janszen), made electrostatic design affordable for loudspeaker consumers.

 

[dlaloum]

Cool looking and congrats on implementing your own design!

Having had Acoustats (20 years ago) I cannot agree that electrostatics are accurate for listening (beamy and not the most accurate frequency response), although they can sound fantastic with just the right listening material, room placement, and if your head is just in the right place.

The beamyness is very dependent on the individual design - some are more "directive" than others.

I found the ESL57 to be quite "beamy" - but the ESL63 & 989 weren't - so it isn't a compulsory feature of ESL speakers...

 

[dlaloum]

If you search the ML owners forum you will find more than a few posts lamenting panels less than 10 years old losing volume. But you will also find responding posts reporting much older panels that still play fine. So I'm wondering if environmental factors are affecting the diaphragm coating. ML's don't use dust covers so UV light exposure could be a factor--- maybe you should keep your pillow cases on!

My speakers have magnetically attached grills on front and back. I show them in posts with the grills off to display their wire stators but the grills are acoustically transparent so I usually leave them on, even when playing music:

View attachment 294420

With the ESL57... it very very easy to ARC the panel - they are already SPL limited, which leads to the temptation to turn things up, and once they arc it pokes holes in the mylar diaphragm, which leads to further SPL reduction, more turning up, causing more ARC-ing - a vicious cycle of destruction.

Solutions tend to be things like using an amp like the Quad 303 - which is limited to V outputs that won't ARC with the ESL57 .... set up that way, they can last for decades without issues.

Alternative approachs spotted "in the wild" include dual stacked ESL57's (stack them and they gain 3db) - or even the very rare Quad stacked ESL57's (total additional gain 6db) but needless to say, that takes up space, and has further reduced WAF...

In a perfect world, the ESL57's legendary midrange abilities, would make it a perfect HT center speaker.... 4 of them in a stacked array behind an acoustically transparent screen....

 

[Jazzman53]

With the ESL57... it very very easy to ARC the panel - they are already SPL limited, which leads to the temptation to turn things up, and once they arc it pokes holes in the mylar diaphragm, which leads to further SPL reduction, more turning up, causing more ARC-ing - a vicious cycle of destruction.

Solutions tend to be things like using an amp like the Quad 303 - which is limited to V outputs that won't ARC with the ESL57 .... set up that way, they can last for decades without issues.

Alternative approachs spotted "in the wild" include dual stacked ESL57's (stack them and they gain 3db) - or even the very rare Quad stacked ESL57's (total additional gain 6db) but needless to say, that takes up space, and has further reduced WAF...

In a perfect world, the ESL57's legendary midrange abilities, would make it a perfect HT center speaker.... 4 of them in a stacked array behind an acoustically transparent screen....

Most ML models can play significantly louder than the Quad57's so I doubt cranking the up was the issue. The ML panels I've seen with low output showed no evidence of burn thru's or holes in the diaphragm or other signs of arc damage, and replacing the diaphragms restored full output so I concluded the conductive coating had deteriorating over time. ML uses a proprietary coating (a vapor deposition of some sort, I think), whereas the Quad 57's used a more reliable nylon coating. I figure ML's coating process is inconsistent or UV could be causing the coatings to fail, but it doesn't seem to be a widespread problem.

 

[Jazzman53]

The beamyness is very dependent on the individual design - some are more "directive" than others.

I found the ESL57 to be quite "beamy" - but the ESL63 & 989 weren't - so it isn't a compulsory feature of ESL speakers...

No doubt that is true (some ESLs are more beamy than others).

It's not well known in the audiophile community but it's perfectly feasible to configure an ESL to radiate a pattern as wide or as narrow as one chooses, using electrical segmentation.

Electrical segmentation is practically never used in commercial ESLs (or if used at all, only minimally implemented) because it adds complexity and cost that conflicts with the profit motivation.

Electrical segmentation is pretty much limited to DIY wire-stator designs (not feasible for other stator types).

Below are directivity sonograms comparing the radiation patterns of an unsegmented flat panel, an unsegmented 30-degree curved panel, and a segmented flat panel. Not all ESL's are beamy!

Below: Schematic showing how electrical segmentation implemented in a 15-segment wire-stator ESL:

 

[Keith_W]

That directivity response for the segmented flat panel is unbelievable. I have never seen anything that measures as perfectly as that. Unfortunately I can't read the scale. And also they did not show vertical dispersion.

 

[RayDunzl]

Martin Logan reQuest:

Measurements across the couch

Solid green has the mic at the center of the couch.

Both speakers playing, so there is waviness due to phase cancellations.

No drop or rolloff in highest frequencies.

Impulse response shows the different timing as the microphone is closer to or more distant from the reference speaker.

Off-center mic position gives two impulses, one for each speaker.

The reference speaker impulse is always in the middle the other speaker impulse shows up as earlier or later.

For the central mic position, the two impulses overlay each other at the center.

The center of the couch gets a nice flat phase response from the panel. It crosses to the woofer at 180Hz..

Other positions, phase gets lost.

Lines manually offset for clarity.

Group delay shows a similar result:

So, I don't see any "beaminess" affecting the higher frequencies off-center across the couch, just the unavoidable phase problems as the listener moves off-asis.

 

[Jazzman53]

That directivity response for the segmented flat panel is unbelievable. I have never seen anything that measures as perfectly as that. Unfortunately I can't read the scale. And also they did not show vertical dispersion.

Unfortunately I lack the gear, software and know-how to make those measurements myself. Those sonograms are an excerpt from one of the ESL gurus on the DIY Audio Forum.

The schematic is mine, though, and shows my DIY ESL. Even if we assume zero vertical dispersion, the panel is a line source which extends above my head when sitting at my listening position so... how bad can it be

It really is amazing how segmentation it works. If I turn the volume low and place my ear close to the panel, it's easy to discern the highest frequencies radiating exclusively from the panel centerline and progressively falling off toward the panel edges.

Just speculating here (my electronics knowledge is poor):
In addition to the frequency attenuation, I surmise there must also be at least some amount phasing attenuation moving from the panel center toward the edges, because the drive voltage traverses thru the groups and each wire group constitutes a capacitor with a finite charging time, in the circuit. It follows that the charging times would constitute a delay line which would in effect bend the radiated wave front.

From my correspondence with Dr. Rod White who authored the technical paper on how to do this (i.e. "Electrostatic Loudspeaker with Zero-Free Polar Response"); I concluded that more/smaller-width wire groups yields wider dispersion, but there is no advantage to making the wire groups less than about 12mm wide (approximate wavelength of the top octave).

All that I know for sure is that it works, from a listener's perspective. I've built both unsegmented and segmented panels and the audible difference is like night and day.

Another interesting consequence is the panel's frequency response. An unsegmented tall flat panel about 10"-12" wide (typical) exhibits a rising frequency response on my crossover's RTA screen, which I attribute to the combined effects of beaming and the dipole roll-off. To compensate this required significant EQ'ing (attenuation of a big peak extending thru the upper mid range/lower treble region, and boosting up the low end).

In comparison; a segmented panel shows essentially flat response on my RTA screen--from the top all the way down to low end cutoff frequency (the low end cutoff frequency is a required input in the Excel spreadsheet program I use to calculate the segmentation component values, which for my ESL is set at 20Hz above where the panel crosses out to the woofer).

 

[Curvature]

I imagine segmentation also helps reduce panel breakup, which otherwise is found in all electrostats.

 

[Jazzman53]

For you're entertainment:

I mentioned earlier that it would be possible to configure an ESL with switch-selectable wide and narrow dispersion modes. I've actually done this with a panel made from welding rods.

It used basically the same segmentation scheme shown in the earlier post except that I inserted a Soviet military multi-pole rotary switch to change modes. Switch "ON" jumped over the resistors to defeat the segmentation network (narrow mode) and switch "OFF" left the network engaged (wide mode). Narrow mode gave the best imaging with a ridiculously small sweet spot and wide mode was great to have when guests dropped in.

The switch mode feature was fun for a while but the novelty wore off pretty quickly because I had to power down to switch modes (lest arcing destroy the switch), and then I had to re-EQ the panel because switching modes radically changed the frequency response. Because of this I ended up leaving the speaker in wide mode most of the time, even when solo listening.

In all of my subsequent speaker builds, I've omitted the switch-mode feature as impractical.

Here's a link to the panel build which used the switch-mode dispersion feature:

A Segmented Wire Stator ESL with Selectable Wide/Narrow Dispersion

I was so inspired by my friend Ken Seibert's ESL project that I decided to have a go at building a pair of segmented wire stator ESL's mysel...

jazzman-esl-page.blogspot.com

 

[teashea]

Do you owe it to yourself to check out the latest electrostatic speaker designs?

Ahh; YES.

No speaker is perfect but I wish everyone could hear the latest commercial and DIY ESL designs.
I'm sure many would find their sound addictive.

I will discuss both, but first I must confess my bias favoring ESLs, which I've championed for years, and in particular my own DIY design.

In my experience, it's far easier for a DIY'er to build a world-class ESL than a world-class conventional speaker. I don't mean that it's less work (the labor is horrendous)-- just that a good result is more certain because you start with less baggage.

I can list as many cons as pros for ESLs but the pros are really good and the cons are not terribly hard to remedy in a good design. Some manufacturers mitigate the cons quite well and the latest offerings are better than ever.

Pros:
1. The ultra-low moving mass diaphragm is so effectively dampened by the greater air mass it's coupled to, that it simply doesn't *ring.
2. Line-source configuration gives really good imaging.
3. Contrary to popular belief and the commercial offerings, DIY designs can tailor dispersion patterns as desired or even provide switch-selectable wide & narrow dispersion modes (costly but do-able).
4. No crossover-- or a hybrid can use a single crossover down low where it's less audible.
5. Typically excellent/superior coherence, resolution & transient response.

*An ESL diaphragm is remarkably resonance-free across the audio bandwidth except for its single fundamental "drum-head" resonance peak which typically occurs in the bass region at a frequency determined by its tension. This resonance is analogous to a woofer's free-air resonance (f/s) and is likewise undesirable.

Cons:
1. The diaphragm's drum-head resonance is a nasty-loud distortion that's difficult to mask.
2. Anemic bass unless the panel is huge, and drum-head resonance can render it boomy & sluggish.
3. Hybrids add the bass but few achieve a seamless blending of the woofer & stat panel.
4. Narrow dispersion which some listeners find objectionable (i.e. "head-in-a-vise" effect).
5. Capacitive load & impedance as low 1 Ohm in treble band requires a strong, stable amp.
6. Long-term reliability is historically iffy, especially in DIY versions.

Mitigating the drum head resonance:
Acoustat and some other full range designs apply a felt strip or mesh fabric on the rear stator which partially absorbs the resonance energy.

Soundlab uses the drum-head resonance to boost bass output by dividing the diaphragm into multiple different-width sections which break up the single/loud/narrow resonance peak into multiple softer resonances spread over a wider bandwidth (i.e. "distributed resonance"). It's an innovative solution that mitigates the booming effect and boosts bass output, but it's not a perfect solution because the the drum head resonance is by definition a distortion, so using it must include its distortion.

Martin Logan tensions the diaphragm quite high, and likewise uses distributed resonance to enhance the upper bass/lower midrange, so that less EQ'ing is required to offset the panel's dipole roll off and flatten its response curve. Here again, distributed resonance spreads out the drum-head resonance, which necessarily includes its distortion.

Mitigating the narrow dispersion:
Narrow dispersion is not necessarily a flaw. In fact, the Sanders/InnerSound speakers are purposely designed or narrow dispersion because it gives the most precise imaging at their tightly focused sweet spot. Some prefer the narrow sweet spot but it's not so good for entertaining guests, and most speaker designers opt for wider dispersion.

Some designs (ML, Soundlab) curve the stat panel to about 30 degrees of arc, which curves its projected wave-front and widens the sweet spot. The small curvature is still a bit beamy but retains much of the magical imaging.

Some designs use a separate, narrow treble panel to spread the highs.

The Quad 63 and similar designs use separate bass panels and a center mid/treble panel with concentric ring conductors sequentially driven thru an LC delay line to function as a point source projecting a spherical wave front.

Mitigating the load capcitance:
Most designs add low value resistors in front of the transformers which renders the load partially-resistive (although still mostly capacitive).

Reliability:
Most manufactures now produce reliable speakers, although 20 years seems to be the best-case limit (except for Acoustats which are seemingly immortal). I would not advise purchasing a DIY ESL unless you do your homework on the specific design, and you and trust the builder to stand behind it.

The weakest link in any ESL is the diaphragm coating degrading over time. So; a good design goal would make the panel easy to disassemble without damaging the stators, and preferably also leaving the diaphragm intact. If thus accessible; the diaphragm could be easily re-coated or replaced.

The most robust designs use insulated wire stators because of their higher resistance to arcing.
Perf metal stators can be reliable if made by a manufacturer like ML who has mastered the specialized processes required to prevent arcing.

I would no longer build or purchase a DIY ESL with perf metal stators because it's so difficult to perfectly smooth-over the cut metal edges and apply the insulating coatings as required to prevent arcing.

The old Acoustat ESLs are practically bullet proof because they used superbly insulated wire stators and a simple carbon black diaphragm coating that remained stable over time-- these continue to play happily for decades!

So; if you want a reliable ESL; buy or build one similar to the Acoustat design. Or at least with stators than can be easily disassembled to access the diaphragms.

*********************************************************************************************************************

My latest DIY design (Jazzman MkIII):
The drum head resonance: Is simply avoided (not excited) by tensioning the diaphragm to resonate <100Hz, and then crossing the panel out to the woofer at least one octave above the resonance using a steep-sloped digital crossover (250Hz @48db/oct). Hence; the panel is not being energized/excited near the diaphragm's resonance frequency.

Dispersion: Tailoring the dispersion requires choosing either a narrow sweet spot with very precise imaging or a wide sweet spot with less precise imaging or something in-between (physics forces this choice-- you can't have it all).
I chose the latter option and tailored the segmentation scheme accordingly.

The stat panel employs symmetrical multi-segmented wire stators driven sequentially from the panel centerline outward to function as a line source projecting a cylindrical wave-front. This works amazingly well; giving wide, smooth-trending dispersion.

Load/Impedance: The RC line used to curve the wave front effectively linearizes the impedance and renders it predominantly resistive, as only the smaller capacitance of the first two wire groups is reflected back to the amp. This also reduces the degrees of phase rotation for the impedance reflecting back to the amp (very desirable).

Woofer/Panel Integration:
The matching dipole radiation patterns of the excellent OB mounted Eminence KappaLite 3012LF woofer and the stat panel merge seamlessly to sound like a single, coherent driver-- the best integration I've heard in a hybrid ESL.

Reliability:
The wood lattice supported wire stators are less impact-tolerant than perf metal stators and must be handled accordingly. Otherwise; they are highly arc resistant resistant and therefore unlikely to be harmed by even crazy-loud volume levels. Even with no protection circuitry, the electronics interface has also proven to be reliable. Over the past five years I've built seven pairs for myself and audio pals, and none have failed.

The stators can be either bond bonded together or mechanically mated in this design. If mechanically mated; the stators can be easily separated to access the diaphragms, should servicing them ever be needed.

How long will the diaphragm coating last? I can vouch for 5 years so far-- but I'll get back to you on that in about 10 years if I'm blessed to still be here.

Website with build photos & write-ups: http://jazzman-esl-page.blogspot.com/?m=0

Photo of the Jazzman MkIII ESL:
View attachment 294103

Very interesting and thoughtful but in the end I have no interest in actually owning them. There are simply too many defects. Compare these to the excellence of Neumann KH150's.