Electrostatic speakers?

[Osse]

Yesterday I was over at the really high-end Hifi store here close to Gothenburg, and really fell in love with that wall of extremly harmonical sound as the electrostatic speakers produce.

I'm searching the net for info, and it seems like you are able to DIY them, and I would like to start up a discussion here if anybody have any experience in the subject.

Anybody tried, and with what result? Conclusions?
Anyone know about good sites in the subject?
Is it an expensive project?

Let the discussion begin
:grin:

 

[natek]

I have seen them taken apart, my old boss was designing some Panel absorber trap's ( he was an acoustical Engineer) for Martin Logan and could describe what i saw for part's, I know you will need a hefty Step up transformer because they run on high voltage, maybe not the most safe of DIY projects, but would be worth saying you built a pair!

Maybe someone else has some more experience at speaker building around here! I also have a pair of vintage Magneplaners (Magnepan) i could peek at for you if you wanted to know any spec's on those. Sorry not much help but maybe giving this a BUMP will strike more interest!:green:

Best of luck!
~Nate

 

[bcarso]

There's at least one book about DIY electrostats. It's a number of years old and it was in softcover. I have it somewhere in storage, but you might be able to find it by doing some searching with hypothetical titles on bookfinder.com.

I personally find electrostats interesting in principle (low moving mass, fairly low distortion) but problematic in practice (terrible beaming of the straightforward flat ones, and a lot of closely spaced resonances that give a false sense of reality). They are also a bitch to drive, as the impedance is predominantly capacitive, especially on the low Z side of the step up transformer you are typically hooking to.

Harold Beveridge did a clever thing long ago with a sort of horn/lens that coupled the planar output of a panel to a better-dispersed acoustical output.

The other thing that could be fun is a direct tube or HV solid-state amplifier---i.e., eliminate the transformer (these were also pioneered by Beveridge). Needless to say not for the faint of heart or shock-prone. I tried to talk a Canoga Park audiophile into letting me make such an amp, but when he realized I wanted to hook it directly to his precious panels he chickened out. And this was someone who HAD to have non-stock in some way or another on everything, so he was sorely tempted.

Such an amp could be done so that its primary frequency compensation was the panel's capacitive load on a high-Z (e.g., ~current) output.

 

[therecordingart]

Wow...good luck with this one!

 

[bcarso]

There appear to be at least two books: Wagner, Electrostatic Loudspeaker Design and Construction, ISBN 0962419168, which is the one I think I have, and Sanders, Electrostatic Loudspeaker Design Cookbook, ISBN 1882580001.

 

[Svart]

http://uk.youtube.com/watch?v=cEeWtBAE5LY

build one of those...

 

[tv]

A couple of tesla coils?
:green:

 

[s2udio]

Any help ??
http://www.eskimo.com/~billb/esloud/eslhwto.html

 

[bcarso]

[quote author="s2udio"]Any help ??
http://www.eskimo.com/~billb/esloud/eslhwto.html[/quote]

Nice bibliography in that reference. I have the Borwick book and look forward to studying that Baxandall article.

By coincidence, I was treated to some great stories about Peter Baxandall last night while quaffing wine with Laurie Fincham at the LA AES meeting. I remember something about Sean Olive's lecture....

 

[PRR]

> a bitch to drive, as the impedance is predominantly capacitive

A common moving-coil cone speaker is a nearly pure capacitor, +plus+ a resistor and a bass-bump.

The resistor is optional: the coil would work fine if super-conductive (right?). And that is just technical details and cost. Therefore an improved MC driver would also be a honking big capacitor, and a pain to put wideband Power into.

Even with the resistor, electromagnetic loudspeakers ARE "a bitch to drive". Come back to the dawn of speakers. For 40 years after the telephone they were too faint (even tarted-up with huge narrow resonance), or were enormous. The only one which caught-on relied mainly on a new HIGH power low-cost tube: brute-force DRIVE. A type 10 could produce an entire WATT of clean audio, a figure never heard in domestic rooms. Then a simple 6" cone in 20" box gave quite good reproduction, at a cost comparable to a Ford or Chevrolet. The history of useful domestic loud speakers is really the history of affordable audio power amplifiers. Type 10, type 50, type 26L6.

All loudspeakers hit the fact that electric motors are heavier than air. The only pseudo-practical work-around is the acoustic transformer, "horn". A light cone and coil on a large horn can appear very resistive and be relatively high (>25%) efficiency, but only to 1KHz or so.

Sanders had the direct-coupled resistor-loaded electrostat amp. Couple transmitting tubes with opamp drive. Simple, direct, major build. When your load is a capacitor, you have a strong need to know the maximum slew of your signal. Sanders picked 6KHz turnover. If the coupling is "good" at 6KHz, then it is 10% good at 600Hz and 1% good at 60Hz. Not any different from a widerange cone speaker. We just moved our 99% losses from speaker mass-air-coupling to amplifier coupling.

Olson has a passage which "proves" the maximum specific output of an electrostat is very very low, meaning unlikely size for music-level loudness.

There was a guy other than Beveridge who analyzed and built horn-loaded electrostats. Hungarian, Czechoslovakian? Very neat acoustic transformer. Try an old AES Loudspeaker reprint book.

> fell in love with that wall of extremly harmonical sound as the electrostatic speakers produce.

They all sound different. The good ones are impressive compared to cones/domes. But much of that is the newness.... they have flaws, just not the same flaws you know so well. They are easy to DIY. With a lot of practice, you can DIY as good as the commercial ones (all of which have been DIY guys making a few extra to sell).

 

[bcarso]

[quote author="PRR"]>> fell in love with that wall of extremly harmonical sound as the electrostatic speakers produce.

They all sound different. The good ones are impressive compared to cones/domes. But much of that is the newness.... they have flaws, just not the same flaws you know so well. They are easy to DIY. With a lot of practice, you can DIY as good as the commercial ones (all of which have been DIY guys making a few extra to sell).[/quote]

Hearing something different is so often appealing but also so often inaccurate. This is one of my major gripes with so-called digital amps, or other open-loop designs. Yes, they interact with the load and produce frequency response anomalies, and they do sound different. Better? I'd prefer to have my tone controls independent of the speaker load.


EDIT:
PS: Just dragged out Borwick. The Baxandall article as updated by Borwick is a gem. Wish I had more time to study it right now.

 

[pstamler]

[quote author="PRR"]> a bitch to drive, as the impedance is predominantly capacitive

A common moving-coil cone speaker is a nearly pure capacitor, +plus+ a resistor and a bass-bump.

The resistor is optional: the coil would work fine if super-conductive (right?). And that is just technical details and cost. Therefore an improved MC driver would also be a honking big capacitor, and a pain to put wideband Power into. [/quote]

Errrr..you sure you don't mean it's a nearly pure inductor? Big coil o' wire, impedance rising with frequency, and all that?

Peace,
Paul

 

[just.sounds]

In the netherlands there is a quite active group of builders.

on this site there is a list of materials that you need. But it is in Dutch so maybe you can babelfish it otherwise drop me a line i can translate some stuff for you.

edit:

here is a site with much info in english covers most of the info on the dutch site

http://www.audiocircuit.com/index.php?cc=941&material=COA#

another seler of material

http://www.metrum-acoustics.nl/Products.html

 

[Osse]

Great info! Thought there were no interest on this one :razz:

I loved the "lager than life" feeling you could recieve, and the dimension you could recieve with sources in different depths of the picture.

I should do some more research when I have more time...

Thanks!

 

[PRR]

> Errrr..you sure you don't mean it's a nearly pure inductor? Big coil o' wire, impedance rising with frequency, and all that?

No. There's a small inductor, 0.5mH, a minor correction to what I said. Actually dominant above the midrange.

The capacitance runs like 300uFd. It defines the upper slope of the bass resonance impedance bump, but then sinks below the 6 ohms of dead resistance.

It is the electromagnetic reflection (through motor-constant) of the moving system Mass. Caps, like Mass, don't like to move quickly.

We may change speaker stiffness and mechanical damping over wide ranges, to infinitely-limp/free if we ignore some practical details. What mostly matters is Mass/Area. This must be minimized for maximum efficiency (less bitch-to-drive), or sub-optimized for extended bass. Area is a compromise between large-signal output and directivity (and cost and decor). There are no zero-mass cone materials, and the useful cones (roll-paper, pulp/felt, skinned foams) are well established. The coil must balance size, mass, and resistance.... lithium is nasty, aluminum and copper are remarkably similar, and all else is worse.

Looking into the jack: ignoring the bass-bump we see 6R+0.5mH+300uFd. The actual useful load, the AIR, is in there shunting the capacitance. It tends to reflect as 10R-20R, which seems like a nice number. But 300uFd is down to 5 ohms at 100Hz, 1R at 500Hz.... the air load is masked by the big capacitance.

The electrostat has negligible series electrical resistance. Suppose we can get superconductor voice coil. Suppose we leave all else the same (our SC has density and conductance similar to Al/Cu, we keep the same magnet). Now the input of a dynamic speaker, neglecting bass-bump, is 0.5mH+300uFd. Ouch! OK, really 0.5mH+(300uFd||10R), which comes to the same thing in most of the audio range.

(Without that dead resistance, the stray inductance does matter a lot, and we might work to reduce it. Since we can't eliminate it, we might reduce its ill effects by reducing the LC resonant Q, say by adding dead resistance in series with our superconductor.....)

 

[moray james]

http://quadesl.nl/sim/
http://user.tninet.se/~vhw129w/mt_audio_design/
http://www.quadesl.org/
http://www.eraudio.com.au/
http://www.izzy-wizzy.com/audio/spkr.html
http://eslrepair.com/default.aspx
http://www.integracoustics.com/MUG/MUG/tweaks/acoustat/
http://www.quadesl.org.uk/
http://mark.rehorst.com/
http://www.quadesl.com/
http://www.integracoustics.com/MUG/MUG/tweaks/quad/hey_you/rebuild.html
http://user.tninet.se/~vhw129w/mt_audio_design/esl_repair_penketh.htm
http://j-ware.no-ip.com/projects/esl.php?setlang=en
http://esl.hifi.nl/index.htm
http://www.ele.tut.fi/~artoko/audio/speakers/esl.html
http://headwize.com/
http://www.diyaudio.com/forums/forumdisplay.php?s=&forumid=54
http://www.sanderssoundsystems.com/support.htm
http://www.elektrostaten-forum.de/index.php
http://www.indoorspecialties.com/ for diaphragm film that is very thin
http://www.head-case.org/
http://www.head-fi.org/
http://www.high-amp.de/html/frame.html
http://www.bigclive.com/index.htm info on how to build high voltage supply
http://www.mdw.ac.at/I101/iea/tm/scripts/jecklin/material/float.pdf

 

[mhelin]

From Rod Elliotts Sound Products projects:
http://sound.westhost.com/project105.htm

 

[bcarso]

[quote author="PRR"]> Errrr..you sure you don't mean it's a nearly pure inductor? Big coil o' wire, impedance rising with frequency, and all that?

No. There's a small inductor, 0.5mH, a minor correction to what I said. Actually dominant above the midrange.

The capacitance runs like 300uFd. It defines the upper slope of the bass resonance impedance bump, but then sinks below the 6 ohms of dead resistance.

It is the electromagnetic reflection (through motor-constant) of the moving system Mass. Caps, like Mass, don't like to move quickly.

We may change speaker stiffness and mechanical damping over wide ranges, to infinitely-limp/free if we ignore some practical details. What mostly matters is Mass/Area. This must be minimized for maximum efficiency (less bitch-to-drive), or sub-optimized for extended bass. Area is a compromise between large-signal output and directivity (and cost and decor). There are no zero-mass cone materials, and the useful cones (roll-paper, pulp/felt, skinned foams) are well established. The coil must balance size, mass, and resistance.... lithium is nasty, aluminum and copper are remarkably similar, and all else is worse.

Looking into the jack: ignoring the bass-bump we see 6R+0.5mH+300uFd. The actual useful load, the AIR, is in there shunting the capacitance. It tends to reflect as 10R-20R, which seems like a nice number. But 300uFd is down to 5 ohms at 100Hz, 1R at 500Hz.... the air load is masked by the big capacitance.

The electrostat has negligible series electrical resistance. Suppose we can get superconductor voice coil. Suppose we leave all else the same (our SC has density and conductance similar to Al/Cu, we keep the same magnet). Now the input of a dynamic speaker, neglecting bass-bump, is 0.5mH+300uFd. Ouch! OK, really 0.5mH+(300uFd||10R), which comes to the same thing in most of the audio range.

(Without that dead resistance, the stray inductance does matter a lot, and we might work to reduce it. Since we can't eliminate it, we might reduce its ill effects by reducing the LC resonant Q, say by adding dead resistance in series with our superconductor.....)[/quote]

This relevant reference, evidently some months old, appeared in an email piece just now:

http://www.audiodesignline.com/showArticle.jhtml?articleID=207800126

 

[PRR]

> relevant reference

Relevant? Very odd numbers. Isn't 6.5Tm a very small Bl value? This and the 200Hz resonance suggest a tweeter, either very small or very low efficiency.

Note that the actual working load, the AIR, never enters into his calculations. It oughta appear in parallel with Res. That's acceptable, but shows how lost most speaker-work is.

With the light mass and the low Bl, this driver is not showing the usual slope from resonance to midrange.

Also the stray self-inductance is very low-Q, as if wrapped loosely around a bar of solid iron. We'd normally model this as several L-R networks with impedance rise slower than a pure inductance. (Perhaps his tweeter has extra low loss coupled to the stray inductance, which may be why he's needing to re-think the Zobel.)

Tweeters are problems. But our first problem is ample level 100Hz-5KHz, the body of the speech/music. A full-range speaker has a full slope above bass resonance, not that skinny anti-notch pip.

 

[bcarso]

Relevant as far as providing a discussion and a equivalent circuit schematic. It is a bit of an odd driver, and indeed the Q seems awfully silly-high, and there is no mention of the effects of an enclosure.

But the author wants to talk mainly about Zobels and the high-frequency end of the impedance, and what can go wrong with amplifiers if the inductive kickback biases the parasitic SCRs on. So, maybe not so relevant in that connection.

A couple of examples of real drivers at two extremes:

http://www.aurasound.com/public/pdf/NSW1-205-8A.pdf

http://www.aurasound.com/public/pdf/NS10-794-4A.pdf


Note that the 1" one despite its size is still pitched as a ~full-range driver. It does work quite well for a near-field one-piece compact stereo system with a local mono woofer, crossing over around 180Hz.

My point earlier about 'stats being difficult to drive is not so much how large the C is, but what I supposed was how close to a pure capacitance they are, with no benefit to the stabilization of typical voltage-output amplifiers from the series R of the cone loudspeakers.