Electrostatic speakers?

[Jazzman53]

ESLs have come a long way in terms of reliability and performance; especially in the best DIY designs, which need not be compromised for commercial viability.

For example; a symmetrically segmented wire-stator ESL can be exceptionally reliable, and can have a sweet spot as wide as any conventional speaker. In fact; the dispersion pattern can be taylored to be whatever the designer wishes. The much higher labor and materials costs is probably why we don't see symmetrical segmentation fully implemented in commercial ESLs, but it's not technically difficult.

Hybrids that use a low-Q woofer on an open baffle, and a DSP crossover, can have quite seamless bass integration, coherence, and really good dymamics.

It may not even be possible to acheive seamless bass integration in a hybrid ESL with the woofer in a box-- at least I've never been able to make that work, and the commercial hybrids I've heard have not achieved it either.

Being a long-time ESL designer/builder, I concede that I'm probably biased... but I can't see myself ever going back to conventional EM speakers.

 

[Arindal]

I have practical experience with ESL (including CLX, as well as Quads) and have a general sympathy for the concept, as well as getting the point why some people in my circle really love them and would never want anything different. I agree that some aspects like midrange transparency, plausible ambience and reverb tonality are exemplary in the best specimen, and can hardly be achieved by conventional dynamic concepts.

That said, I would like to see more of understanding from ESL haters for their undisputies qualities, as well as concession from fanboys that there are practical limitations beyond sheer size, and some some of the urban legends why ESL are unique, are just that: legends. For example the myth about low diaphragm mass and resulting perfect transient response. Sorry, but this contradicts basic laws of acoustics, in my understanding.

Shall I continue?

What about:

1. overly broad and tall phantom source localization due to sheer panel size
2. tendency to get rather diffuse localization and ambience
3. very restricted horizontal listening window / sweet spot
4. Limited SPL capabilities in the bass and lower midrange bands
5. Tendency of overly reduced subjective proximity due to early rear wall reflections (same problem like with any true dipole)

Everyone is free to accept such shortcomings, as every speaker on this planet is a compromise. But we should not discuss ESL as the one and only concept having no downsides at all. That is simply not true.

a symmetrically segmented wire-stator ESL can be exceptionally reliable, and can have a sweet spot as wide as any conventional speaker.

I seriously doubt that. Could you please explain how frequencies above 10K can be equally distributed to a +-50deg window (horizontally) from a diaphragm which is broader than the wavelength? How slim do you expect the active area to be in order to achieve this?

but I can't see myself ever going back to conventional EM speakers.

Why not? Which single aspect of sound quality would you consider being impossible to achieve with conventional transducers?

Having studied a lot of planar/dipole speaker concepts, my personal conclusion would be that one could achieve the major advantages of ESL with conventional transducers as well. I don´t see a secret magic in ESL, making them unique. Would presume it is just a matter of achieving similar directivity, wave emanation (line source / planar source / dipole hybrid), avoiding resonances as well as cancellation and edge diffraction issues, while using similar transducer principles for the treble (like planar magnetostats). I have heard in a video by Paul McGowan that he and Chris Brunhaver had a very similar ideal when designing their PS Audio speakers. Same with Dieter Fricke of Ecouton, whose fullrange dipole concept brought me to be thinking that such concept would work:

Or is anyone seeing an acoustic reason why this is impossible?

 

[Jazzman53]

The diaphragm in my ESL speaker is 10.5" wide x 46.5" tall. However, only a narrow portion of the diaphragm emits treble.

Specifically; my stators' (15) driving wire groups are electrically segmented such that only a 0.5" wide x 46.5" tall zone in the center of the diaphragm radiates frequencies above 10kHz. As symmetrically segmented, the panel functions as a line source projecting a cylindrical wavefront. The details are given on my website here: https://jazzman-esl-page.blogspot.com/

Conventional EM speakers can sound pretty amazing but for me, the clarity and coherence of the best ESL designs seal the deal. Besides; I wouldn't buy any commercial speaker, even an ESL, when I can build my own and control every design feature.

 

[Arindal]

my stators' (15) driving wire groups are electrically segmented such that only a 0.5" wide x 46.5" tall zone in the center of the diaphragm radiates frequencies above 10kHz.

0.5" width of the active zone sounds like a very good idea for upper treble and should contribute to a wider pattern indeed. I am a bit reserved when it comes to overly tall tweeters or tweeter zones, as this oftentimes results in vertically blurred or stretched localization. An active zone like 0.5" x 8" for the treble would be ideal. Same with dipole characteristics resulting from slim tweeters in what appears to be a very broad baffle or midbass planar as they are oftentimes not dipoles but more diffus bipoles.

Nevertheless very impressive build, not saying that this would not work!

Conventional EM speakers can sound pretty amazing but for me, the clarity and coherence of the best ESL designs seal the deal.

There was a time when I thought so, too. Only after discovering dynamic speakers which over the whole midrange resemble ESLs in terms of directivity (dipoles, cardioids, line sources or a combination thereof) and active transducer area (de facto point/line sources) as well as similar characteristics in the upper treble (planar magnetostats or AMTs), I changed my mind. Admittingly, these are rare, the Ecouton I have mentioned is not very popular.

 

[misterdog]

At the end of the day audio reproduction is merely an illusion, in the same way that TV or cinema is only 2 dimensional, trying to represent a 3 dimensional reality.

At the end of the day my ESL's bring me closer (IMO) into that illusion than any cone based speaker I have yet heard.

I have friends who have spent 20/30 years perfecting their cone speaker systems, yet all I hear is

Box colouration.
Crossover distortion.
So much energy being pushed from the box that it excites room nodes.

Despite the supposed elitism of planar speakers due to size and cost I'll stick with what makes my ears most comfortable over a long period of time.

 

[misterdog]

Everyone is free to accept such shortcomings, as every speaker on this planet is a compromise. But we should not discuss ESL as the one and only concept having no downsides at all. That is simply not true.

Has anyone ever suggested that ESL's or planar speakers have no downsides?

From the attached chart which 'real analogue' instruments are ESL's failing to reproduce?
The failings of ESL's at frequency extremes being the most common criticism.

Why are there so few, if any, musical instruments made which can reach the limits of (perfect) human hearing?

 

[Purité Audio]

At the end of the day audio reproduction is merely an illusion, in the same way that TV or cinema is only 2 dimensional, trying to represent a 3 dimensional reality.

At the end of the day my ESL's bring me closer (IMO) into that illusion than any cone based speaker I have yet heard.

I have friends who have spent 20/30 years perfecting their cone speaker systems, yet all I hear is

Box colouration.
Crossover distortion.
So much energy being pushed from the box that it excites room nodes.

Despite the supposed elitism of planar speakers due to size and cost I'll stick with what makes my ears most comfortable over a long period of time.

Your friends must have really crappy speakers.
Keith

 

[Pearljam5000]

I liked electrostatics ever since I saw them on Friends as a kid

 

[Newman]

0.5" width of the active zone sounds like a very good idea for upper treble and should contribute to a wider pattern indeed.

It sure would, but it is not what is really happening. As you can see from the following theoretical beam pattern of a segmented ESL panel, it is nothing like the pattern of a 0.5" wide line source.

A 0.5" wide line source would be constant width right out to 20 kHz. The above is not.

As I have previously mentioned, and tagged you in, it would be great if we had a competent and comprehensive independent trustworthy measurement of the actual output from a real, complete segmented ESL speaker. But I have yet to see one. I have a suspicion that, when we get one, there will be howls of disappointment, machine-gun-like shooting of the messenger, accusations of the measurement not being suited to the speaker, the whole caboodle. Just like happened when Amir tested a small Magnepan. Yes, denial is predictable.

...There was a time when I thought so, too [that ESL clarity and coherence is superior]. Only after discovering dynamic speakers which over the whole midrange resemble ESLs in terms of directivity (dipoles, cardioids, line sources or a combination thereof) and active transducer area (de facto point/line sources) as well as similar characteristics in the upper treble (planar magnetostats or AMTs), I changed my mind. Admittingly, these are rare, the Ecouton I have mentioned is not very popular.

Agreed. ESLs are a prime example of the Exotic Speaker Myth: that having some theoretical special aspect of performance must translate into a superior experience in the listening room. Once that notion takes hold, confirmation bias tends to run rampant. I have previously posted on some of the ESL Myths and why they take hold.

cheers

 

[Jazzman53]

Just to clarify;
The simulated directivity sonogram shown on my website and/or the one shown in the previous post, is generic and likely not based on a 0.5" wide line source, but for a much wider symmetrically-segmented panel where the entire panel radiates the lower frequencies, and the upper frequences are progressively/sequentially attenuated from the center segment to the outer segments arrayed on either side.

Many or most of us tend to ignore anectdotal/subjective descriptions not supported by actual measurements, and that's understandable given the pervasive hype in this hobby. I'm not equipped to measure and show my panel's dispersion pattern, so I can only offer my subjective listening impressions.

I trust the math and my own ears though, and my subjective assertion is that symmetrical segmentation done right works as theorised, and amazingly well. By "done right" I mean a finely segmented panel with a sufficient number of narrow (12mm) segments to produce a smooth-trending dispersion pattern.

There are a couple of commercial wire-stator ESLs that employ a separate treble segment at the inside edge of the panel (an assymetric 2-segment panel) or at most, separate bass, mid and treble segments (asymmetric 3-segment panel). Those manufacturers consider two or three segments to be 'good enough' insofar as it mostly alleviates the "head in a vice" effect. However, just two or three segments can't produce the wide, smooth-trending dispersion of a 15-segment symmetrical panel (7+1+7 symmetry with a single center segment and seven left & right paired segments arrayed on either side).

To each his own of course, and I'm happy with what I have until I hear something I like better.

Charlie

 

[Keith_W]

Just like happened when Amir tested a small Magnepan. Yes, denial is predictable.

You talking about Amir's Magnepan LRS review? I wouldn't make conclusions on all ESL's based on measurements of one. Amir had a lot of problems measuring it, and this particular speaker had very little treble or bass extension - an inverted U-curve.

What I learned from his review is that planar speakers are very difficult to measure, even with a Klippel. Amir noted that the interaction between front/rear drivers produced a complex sound field at short wavelengths. Since we have @NTK in this thread (a sound field nerd) it makes me wonder if the maths used by Klippel to extrapolate the far-field from a near-field measurement somehow fails with planar speakers, and the fact that this happens at high frequencies might be a clue. Anyway since you mention it, I should go re-read that thread to see if it has already been discussed.

 

[RexrothPigeon]

I trust the math and my own ears

you may be right, I don't know (and don't really have strong feelings either way), but if there's one thing I've learned here on ASR it's that unless we are listening blind, under carefully controlled conditions, we are not trusting our own ears because our ears are overruled by our eyes and unconscious biases and preconceptions. This phenomenon is so powerful that it persists even when we are wary of it and making great efforts to resist it, and it applies to every last one of us.

 

[Newman]

You talking about Amir's Magnepan LRS review? …Anyway since you mention it, I should go re-read that thread to see if it has already been discussed.

Also check out the thread that Amir opened to discuss his intentions to measure that speaker. That alone attracted nearly 120 posts with all the usual hand waving.

 

[MRC01]

You talking about Amir's Magnepan LRS review? I wouldn't make conclusions on all ESL's based on measurements of one. Amir had a lot of problems measuring it, and this particular speaker had very little treble or bass extension - an inverted U-curve.

Also, the LRS that Amir measured does not have the 5' long ribbon tweeters that the bigger models have, which have more even dispersion and lower distortion. Those tweeters are excellent and a big part of what built Magnepan's reputation over the years.

What I learned from his review is that planar speakers are very difficult to measure, even with a Klippel. Amir noted that the interaction between front/rear drivers produced a complex sound field at short wavelengths. Since we have @NTK in this thread (a sound field nerd) it makes me wonder if the maths used by Klippel to extrapolate the far-field from a near-field measurement somehow fails with planar speakers, and the fact that this happens at high frequencies might be a clue. Anyway since you mention it, I should go re-read that thread to see if it has already been discussed.

Yes, big planars are hard to measure. Near field isn't what you hear, and far field includes the room as much as it does the speaker. The accuracy of the Klippel simulation/extrapolation was discussed in that thread, for example here: https://www.audiosciencereview.com/...magnepan-lrs-speaker-review.16068/post-527691

 

[NTK]

You talking about Amir's Magnepan LRS review? I wouldn't make conclusions on all ESL's based on measurements of one. Amir had a lot of problems measuring it, and this particular speaker had very little treble or bass extension - an inverted U-curve.

What I learned from his review is that planar speakers are very difficult to measure, even with a Klippel. Amir noted that the interaction between front/rear drivers produced a complex sound field at short wavelengths. Since we have @NTK in this thread (a sound field nerd) it makes me wonder if the maths used by Klippel to extrapolate the far-field from a near-field measurement somehow fails with planar speakers, and the fact that this happens at high frequencies might be a clue. Anyway since you mention it, I should go re-read that thread to see if it has already been discussed.

The limitation with the Klippel NFS is that the choice of using spherical harmonics as the expansion functions means that the math is not efficient for sound sources that aren't compact (compact means small when compared to the wavelength of sound being produced). To "fit" the radiation pattern of a non-compact source like a panel speakers (or a line array) will require an enormous number of expansion functions, and therefore an enormous number of measurement points. (The number goes up fast once the driver size to wavelength ratio exceeds a certain limit, but I am not sure how fast. Power of 2? 3? Exponential?)

It is kind of analogous to the Fourier series, which isn't very efficient in approximating a square wave, and requires a big long series of sine functions to give a good approximation. (For Fourier series, if you have an impulse type discontinuity, the magnitude of the Fourier coefficients stay constant with frequency and the series doesn't converge. If you have step type discontinuities, as in a square wave, the Fourier coefficient magnitudes decrease linearly with frequency and the series converges slowly. For triangular waves, they decrease with the square of frequency and the series converges faster. Etc.)

However, if one uses the Haar wavelet (i.e. uses the Haar sequence as the basis functions instead of sines and cosines), a square wave is can be "efficiently" expanded. But the Haar wavelet won't "efficiently" fit a sine wave, whereas Fourier does.

Typical speakers with dynamic drivers are usually compact sound sources as only one driver is active most of the time, and when at cross-over regions, the driver spacing to wavelength ratio still isn't too bad. So 1000 points or so is usually sufficient. In the case of the LRS, the estimated fitting errors exceeded the default limit of 1% at ~ 3 kHz.

As for in-room effects, I doubt they are any different from other type of speakers. Panel speakers have different radiation patterns compared to the typical forward firing ones with dynamic drivers. But IME the acoustics and psychoacoustics work the same, i.e. "resonances" (i.e. uneven frequency responses, on or off axis) are bad either way.

 

[Keith_W]

See, I told you he's a sound field nerd

 

[Newman]

See, I told you he's a sound field nerd

Although I don't think NTK's post justifies your descriptions in bold, below.

[to Newman] ...I wouldn't make conclusions on all ESL's based on measurements of one. Amir had a lot of problems measuring it, and this particular speaker had very little treble or bass extension - an inverted U-curve.

What I learned from his review is that planar speakers are very difficult to measure, even with a Klippel.

Okay, now you are the one who is "making conclusions on all ESLs based on the measurements of one (non-ESL planar)."

I don't think Amir had any problems measuring it. He just needed more time than he was willing to put in to get super-accurate in the top two octaves.

Amir noted that the interaction between front/rear drivers produced a complex sound field at short wavelengths.

I don't recall any such note about front/rear interaction. Amir said, "In theory, this should NOT have happened. Lay intuition about this speaker is that they are two vertical drivers each sending out a perfect plane wave of their own. Reality is different in that there are clearly other sources of sound interfering with each other, creating a highly complex wave front."

I read this as saying that a well-behaved line source would have been much more straightforward, and that the LRS is poorly behaved. The issue isn't that all planars or dipoles are creating complex sound fields, rather, the LRS is messy.

Since we have @NTK in this thread (a sound field nerd) it makes me wonder if the maths used by Klippel to extrapolate the far-field from a near-field measurement somehow fails with planar speakers, and the fact that this happens at high frequencies might be a clue. Anyway since you mention it, I should go re-read that thread to see if it has already been discussed.

No I think, albeit as a non-nerd, the messaging from Klippel/Amir is more like this: it can model any sound source of any complexity from its measurements, given enough time to produce many measurement points and grind through the processing and large files. One can think of the imaginary conversation between Amir and his machine a bit like this:-
Amir: "Measure this, will ya?"
Herr Klippel: "Can do. But precision is going to take this amount of time for so many measurements and mucho processing, and result in files of this size. Okay?"
Amir: "Errr, no. Just do 2000 points and we'll take it as it comes."
Herr Klippel: "Here ya go, Mr Compromiser."

I really don't think you are on the right track to allude that it is "failing" with panels/dipoles, and especially not to hint that the errors (uncertainty levels) in the high treble from a 2000 point measurement suggest that there could be some overall failure.

cheers

PS here is Amir's final summary comment on the LRS review, after more than 1500 replies: "The only limitation was the response of upper treble that could have been remedied with more measurement points. Klippel NFS treats all speakers as sources of sound and doesn't care or make any assumptions about how they do that. It samples radiated sound in 3D space and solves the wave equations for it."

 

[dlaloum]

"midrange" is more of a marketing word than a technical one as there is no specific definition of "midrange". It's generally regarded as the range of human voices, but a LOT of instruments operate in that range, too.
Quad ESL-63s will go down to around 80 Hz and up to about 15 kHz with reasonably flat in-room response. Most electrostatics, except the narrower ones, will do about the same.

As for difficulty in placement... it depends on the room, as it does with any speaker. The main difficulty is the size- they tend to optically dominate a small room. Since they are usually rather directive, at least at higher frequencies, you can easily control where the direct signal and reflections will go.

There are a lot of people who claim that you can't properly integrate a conventional bass driver in a box with ESLs. Based on my own experience, this is nonsense. My ESL-63s with a single SVS 3000 Micro sub sound fantastic, and you'd never know the lowest octave or two wasn't coming from the Quads.

It was very difficult to integrate a sub with ESL's right up until DSP integration software became common...

Nowadays it is not so much of an issue!

 

[dlaloum]

At the end of the day audio reproduction is merely an illusion, in the same way that TV or cinema is only 2 dimensional, trying to represent a 3 dimensional reality.

At the end of the day my ESL's bring me closer (IMO) into that illusion than any cone based speaker I have yet heard.

I have friends who have spent 20/30 years perfecting their cone speaker systems, yet all I hear is

Box colouration.
Crossover distortion.
So much energy being pushed from the box that it excites room nodes.

Despite the supposed elitism of planar speakers due to size and cost I'll stick with what makes my ears most comfortable over a long period of time.

From my listening observations... dynamic speakers that don't use a box, and instead use other shapse (cylinders, spheres, etc...) do not seem to suffer from the dreaded "box sound" that those of us who have spent years listening to ESL's (or various other panel speakers) tend to be sensitive to.

The Gallo spheres do a damn good job of sounding very much like an ESL - without the boxiness.... the Ref 3.2 I have has a cylinder for the woofer, and spheres for the midranges - no large flat surfaces to radiate nasties and the solid spherical/cylindrical metal enclosures are very resonance resistant.

Point being, I don't think it is a problem related to the nature of the drivers, rather it is the nature of the enclosure structure and sound signatures related to the way the most common enclosures (boxes) resonate.

Sadly, non-boxes (whether panels, or spheres/cylinders/other) continue to be a niche category....

 

[Arindal]

Has anyone ever suggested that ESL's or planar speakers have no downsides?

Maybe not explicitly, but their supporters oftentimes admit rather secondary downsides (like size or sensitivity), rather than discussing clear and inevitable degradation of sound quality. In my understanding that leads to unnecessary hostile positions from both extreme positions, facilitating what occasionally looks like a flamewar between fundamentalists, rather than a serious discussion on advantages and disadvantages.

From the attached chart which 'real analogue' instruments are ESL's failing to reproduce?

Every instrument that in order to be perceived as realistic, requires very sharp localization (including the human voice and everything that is intentionally close-mic´d), a certain amount of proximity and ´transient attack´ as well as (lower) bass at sufficient SPL, will either this or that way pose a compromise. We should not forget that a lot of recorded music does not consist of ´real analogue´ instruments, but electronic sounds, and people have different expectations in terms of SPL and bass.

You are naming certain downsides, which admittingly many conventional speakers show (boomy bass due to excited room modes, ´box coloration´ as a result of baffle step and resonances), as if they were inherent to dynamic boxed speakers. They are not, they just require more of effort to be overcome, but I would say this is possible without the downsides of ESL which are inherent to the concept.

As you can see from the following theoretical beam pattern of a segmented ESL panel, it is nothing like the pattern of a 0.5" wide line source.

I agree, but calculated from the ´diaphragm step´ setting in above 2K, it becomes clear this active zone is broader than 0.5".

ESLs are a prime example of the Exotic Speaker Myth: that having some theoretical special aspect of performance must translate into a superior experience in the listening room. Once that notion takes hold, confirmation bias tends to run rampant.

I agree that ESL are oftentimes referred to as an ´anti-mainstream´ concept (as are horns, omnis, fullrange cones), which might bring their supporters to thinking they are always superior. I consider myself completely unbiased here and have been conducting blind tests involving ESL quite some time again, but found at least some aspects of sound quality which at the time were pretty unique to ESL (like midrange directivity, transparency, balanced reverb tonality, bass impulse precision for fullrange dipole planars). So I would conclude this is beyond the usual ´anti-mainstream´ bias and should be recognized, without leaving the downsides of ESL unmentioned.

The issue isn't that all planars or dipoles are creating complex sound fields, rather, the LRS is messy.

I agree, and according to my experience with having planar dipoles in the anechoic chamber, I would say this is mainly due to diaphragm breakup, out-of-phase behavior, resonances and de-facto bending wave properties. This seems to be inherent to large area planars or overly tall, non-suspended ribbons, and I see it as a reason why planars like Magnepan, Apogee and alike in a room never really sound like the measurements would suggest. This is not a Klippel problem, btw.

dynamic speakers that don't use a box, and instead use other shapse (cylinders, spheres, etc...) do not seem to suffer from the dreaded "box sound" that those of us who have spent years listening to ESL's (or various other panel speakers) tend to be sensitive to.

This would hint to the explanation that internal resonances and edge diffraction issues are the main reasons for the ´dreaded box sound´ as you define it.

We should not that such enclosure and baffle shape does not take away the baffle step and directivity issues of conventional speakers, which I see as a main reason why ESL in many cases sound different (better, more natural, plausible ambience) in the midrange.