Anatomy of a high efficiency high output electrostatic loadspeaker.

MartinLogan Audio Owners Forum

Help Support MartinLogan Audio Owners Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I did a search and found this ideal coating material, very much like was used in DW speakers. I calculated that the charge diffusion time constant should be about 100ms. This would is roughly 5 times the period of 50 Hz. If you use the panel at a higher frequency you can chose a shorter time constant. Given that the stator to cell capacitance is 120pf, the resistivity needs to be about 850 Mohm per square.
So this liquid will fit the bill and you get a big quantity in each can. But it is even better! you need to dilute this 5 times to get the high resistivity and now you have 4L of liquid to share with all your friends. The instructions are clear, clean with isopropyl alcohol, this will not only clean, but slightly etch the membrane to make the solution stick for decades. I believe I got a sample of this years ago. The person on the phone told me that is what Beverege uses on their speakers. Remember those days when you actually called up someone and they helped you out? MG Chemicals 844AR ESD coating, 850mL.. No reason to go and find some fancy ESL builder that sells you a few ounces for a high price.

https://www.amazon.com/MG-Chemicals...f7644529fb7430a39e658c8d1fe70801&gad_source=1
 
I had another observation about ESL construction. I watched the video on ML construction. I see that the diaphragm is sandwiched between two layers of tape. How can that be good structurally? The top and bottom stators could be flopping on transients? There are other speaker builders that use tape instead of glue and screw/rivet. Rivets are used in airplanes so that the vibration does not loosen the screws. So can anyone explain to me why not a plexiglass or delrin {probably too expensive) spacer for gap and diaphragm glued to one side and other stator mechanically fastened? I am rebuilding a turntable and it is extremely important to retrieve detail by connecting the tonearm as rigidly to the spindle of the platter.
So here we have the stators flopping around under transients because of the give in the tape. Anybody actually made a cell both ways to see if the rigidity can be heard on transients? As soon as I get another working panel and finally get both channels of my ML speaker working, I can listen to all my old material. The kind of stuff I am taking about are rim shots, wooden blocks, tight guitar, and any type of percussion. If the reproducer is outstanding, you can hear the air around each instrument. But if your system has a lot of mechanical hysteresis, and almost all the cone speakers I have heard do, you lose some of that air. This has nothing to do with room bounce, my headphones reproduce it and there is no room bounce there.
An example of the best reproduction of resolving transients are the chimes on the album Dark Side of the Moon. When i first heard it on an high fidelity system I thought this sounds exactly like a real grandfather clock like my grandfather had.
 
I had another observation about ESL construction. I watched the video on ML construction. I see that the diaphragm is sandwiched between two layers of tape. How can that be good structurally? The top and bottom stators could be flopping on transients? There are other speaker builders that use tape instead of glue and screw/rivet. Rivets are used in airplanes so that the vibration does not loosen the screws. So can anyone explain to me why not a plexiglass or delrin {probably too expensive) spacer for gap and diaphragm glued to one side and other stator mechanically fastened? I am rebuilding a turntable and it is extremely important to retrieve detail by connecting the tonearm as rigidly to the spindle of the platter.
So here we have the stators flopping around under transients because of the give in the tape. Anybody actually made a cell both ways to see if the rigidity can be heard on transients? As soon as I get another working panel and finally get both channels of my ML speaker working, I can listen to all my old material. The kind of stuff I am taking about are rim shots, wooden blocks, tight guitar, and any type of percussion. If the reproducer is outstanding, you can hear the air around each instrument. But if your system has a lot of mechanical hysteresis, and almost all the cone speakers I have heard do, you lose some of that air. This has nothing to do with room bounce, my headphones reproduce it and there is no room bounce there.
An example of the best reproduction of resolving transients are the chimes on the album Dark Side of the Moon. When i first heard it on an high fidelity system I thought this sounds exactly like a real grandfather clock like my grandfather had.

Some years ago I gave serious thought to starting my own speaker company but I talked myself out of it after realizing how much I would have to simplify and compromise my [very labor-intensive] design to make it commercially viable.

I suspect the ML founders faced a similar dilemma, and made the best choices they could to move forward. Compromises are often necessary for marketing but a product still has to satisfy the end users, at its price point.

The double-sided urethane foam tape itself cost more than a hard shim would, but it provides huge labor savings because it bonds the panel details together instantly, with minimal fuss, without fasteners, glue or glue set-time, and its compressibility accommodates more fit error than a hard shim ever could.

Having recently rebuilt my first [and last] ML panel, and having no prior experience with ML's, I was surprised at how much non-uniformity I measured in the tape thicknesses, and the asymmetrical diaphragm-to-stator spacing (d/s). The spacing asymmetry results from the front spacer tape being about 0.025 thick, and the rear tape spacers being about 0.045 thick.

The diaphragm in an ML must be tensioned over a curved stator, causing it to become "saddle-shaped" between the spars, and closer to the rear stator in the center area between the spars.
It makes sense then, that the rear spacers must be thicker to accommodate the "saddles" in the diaphragm.

Additionally; the curvature imparts a hoop resistance acting to prevent the diaphragm from moving forward as easily as moves backward-- a compromise imposed by the curved panel.

At least the curvature gives substantial rigidity to the stators/panel, and in the panel rebuilt, I came to appreciate the compromises necessary to curve the panels for wider dispersion.

In a perfect world I would use super-dense tungsten wire for the stators, with the diaphragm and panel hard coupled to a massively dense frame, so as to broadcast maximal sound energy with minimal energy lost to vibrating the stator and frame.

If I had ever heard such a perfect speaker, I might then be less satisfied with the less than perfect ESLs I've heard.

My own DIY speakers use tape spacers, and its wood-lattice supported wire stators vibrate big-time, yet they still sound marvelous (until I hear a perfect ESL).

ML panels likewise vibrate significantly, and even with all the asymmetries and compromises described above, they still sound pretty amazing (same caveat).
 
Last edited:
The option chosen by DW was to use multiple panels curved into an array, so now we have labour for 8 on the original and 10 on the newer version. You now need two sets of frames to arrange the 10 panels and with gas you need an outer frame to hold that and another set of front and back diaphragms.
When I replaced the what can be called a bag, you remove the original plastic from the metal frame using a sander to remove all the plastic. Then there is a two part glue that is pretty smelly and I am sure is no longer sold to home users. The two part is because of an accelerator to harden the glue, the same way epoxy works. You got a 30 minute working time to place the 3 ft by 3 ft sheet on the frame before the glue sets. I probably have some photos not scanned in of that procedure. Then the glue reaches full strength after an hour and you let loos the hair dryer on the Mylar to tension out the wrinkles. I can't image using any other Mylar you somehow have to stretch and hold before the glue dries. Also the reason I did this again, was that I used epoxy thinking that it would work. Epoxy does not work gluing Mylar to a steel frame.
If you run the speaker without gas, and hold you hand on the inner support while playing loud music, you realize the large forces at play to move that thin piece of plastic to move the air.
And yes, the tape is convenient but having different front to back spacing means significant second harmonic distortion.
DW published distortion numbers for the speakers. I known there are none for the ones i have. Do they publish the numbers for their flagship speaker? I just looked at the ML Statement. No specs.

@Jazzman52, did you ever measure distortion on your speakers? Its easy to do nowadays with a not expensive Berhinger microphone that has a good flat response, free soundcard software and of course a soundcard. No fancy lab equipment needed today. Statements like the speakers sound good is not a scientific definition. Now if someone said, well is it Real or is it Memorex, Oops, my Speaker.
 
The option chosen by DW was to use multiple panels curved into an array, so now we have labour for 8 on the original and 10 on the newer version. You now need two sets of frames to arrange the 10 panels and with gas you need an outer frame to hold that and another set of front and back diaphragms.
When I replaced the what can be called a bag, you remove the original plastic from the metal frame using a sander to remove all the plastic. Then there is a two part glue that is pretty smelly and I am sure is no longer sold to home users. The two part is because of an accelerator to harden the glue, the same way epoxy works. You got a 30 minute working time to place the 3 ft by 3 ft sheet on the frame before the glue sets. I probably have some photos not scanned in of that procedure. Then the glue reaches full strength after an hour and you let loos the hair dryer on the Mylar to tension out the wrinkles. I can't image using any other Mylar you somehow have to stretch and hold before the glue dries. Also the reason I did this again, was that I used epoxy thinking that it would work. Epoxy does not work gluing Mylar to a steel frame.
If you run the speaker without gas, and hold you hand on the inner support while playing loud music, you realize the large forces at play to move that thin piece of plastic to move the air.
And yes, the tape is convenient but having different front to back spacing means significant second harmonic distortion.
DW published distortion numbers for the speakers. I known there are none for the ones i have. Do they publish the numbers for their flagship speaker? I just looked at the ML Statement. No specs.

@Jazzman52, did you ever measure distortion on your speakers? Its easy to do nowadays with a not expensive Berhinger microphone that has a good flat response, free soundcard software and of course a soundcard. No fancy lab equipment needed today. Statements like the speakers sound good is not a scientific definition. Now if someone said, well is it Real or is it Memorex, Oops, my Speaker.
I've never measured the distortion level of my speakers.

I do have a mic, and I once attempted to setup REW but gave up on that idea because it would have required inserting an analog preamp to run the REW thru, and I didn't want to do that because I don't play music thru an analog preamp. Rather; I use the digital preamp in my Logitech Transporter streamer and there is no way to run the REW connections thru the Transporter.

BTW; my speakers have flat panels with symmetric/same spacer thicknesses front & back. I also suspect that my round-wire stators should rebound less energy back to the diaphragm, compared to the flat inner surfaces of perf-metal stators. It would be interesting to measure and verify that.

And oh BTW;
I've built seven pairs of the Jazzman MkIII speakers.
Four pairs were built for my audio friends at the CarverSite forum, 1 pair are mine, and I'm keeping one pair for swap out in case I ever need to honor the warranty on any of the pairs I've sold.

That leaves one unused/unsold pair, including magnetic grills front & back, that I would sell at a bargain price, which includes a 30-day full refund satisfaction guarantee backed up with a five-year industry standard parts & labor warranty (5 years or my lifetime, whichever comes first).

However, buyer would have to pick them up in Savannah, GA. My speakers were not designed to sold or shipped, they don]t fit standard shipping containers, and so shipping would require special crates and the cost to ship would be prohibitive.
 
Last edited:
I am really looking forward to hearing therm when I get a chance. Most likely in within the next 4 months timeframe.\

My plan now is to measure the distortion of the ML stepup transformer and the DW as soon as I build an HV test probe. I got too many electronic projects on the go. But I have all the parts for the probe. During my design plan for a new arrangement of panels I purchased the minimum lot required of 200Mohm 15kV resitors. I also have 1Mohm 15kV resistors. I once tried to measure the 15kV bias voltage with stacked resistors, only to burn out my digital meter. Now I can stack 5 200 Mohm ones to get 1000:1 HV probe. Probably silicon the resistors on to a thin strip of plexiglass. Those resistors cost me $220 in 1999. I think i have 150 of them. I don't remember what they were for now.
 
I am really looking forward to hearing therm when I get a chance. Most likely in within the next 4 months timeframe.\

My plan now is to measure the distortion of the ML stepup transformer and the DW as soon as I build an HV test probe. I got too many electronic projects on the go. But I have all the parts for the probe. During my design plan for a new arrangement of panels I purchased the minimum lot required of 200Mohm 15kV resitors. I also have 1Mohm 15kV resistors. I once tried to measure the 15kV bias voltage with stacked resistors, only to burn out my digital meter. Now I can stack 5 200 Mohm ones to get 1000:1 HV probe. Probably silicon the resistors on to a thin strip of plexiglass. Those resistors cost me $220 in 1999. I think i have 150 of them. I don't remember what they were for now.
So you'll be feeding into a distortion analyzer? Careful not to burn that out!

I've used a standard 1000:1 hv probe to measure the bias of my CLS II's. But since the impedance of the bias supply is so high, I don't consider the measurement reliable. Only for comparison shopping, when I wasn't convinced one of the bias supplies was turning on with a music signal.
 
I do have a mic, and I once attempted to setup REW but gave up on that idea because it would have required inserting an analog preamp to run the REW thru, and I didn't want to do that because I don't play music thru an analog preamp. Rather; I use the digital preamp in my Logitech Transporter streamer and there is no way to run the REW connections thru the Transporter.
Doesn't the Logitech Transporter have a USB in? Most DACs today are recognized in Windows or Linux (and probably MAC OS but I have no experience) as a sound card. These days, I run my computer's HDMI output into my Emotiva a/v proc to use REW. I also listen to analog sources, all the time, through a separate analog preamp, but using the analog output of a computer would generally not produce good results.

I looked at the Transporter when I was shopping for DACs years ago (having already amassed a fleet of Squeezebox players), but went with the Benchmark DAC1 Pre instead (which definitely can act as a sound card). The deal breaker for me was it didn't have a discrete remote command for inputs, only to toggle from one input to the next (am I right?) so programming an "activity" in a universal remote would have been kludgey or impossible.
 
I had another observation about ESL construction. I watched the video on ML construction. I see that the diaphragm is sandwiched between two layers of tape. How can that be good structurally? The top and bottom stators could be flopping on transients? There are other speaker builders that use tape instead of glue and screw/rivet. Rivets are used in airplanes so that the vibration does not loosen the screws. So can anyone explain to me why not a plexiglass or delrin {probably too expensive) spacer for gap and diaphragm glued to one side and other stator mechanically fastened? I am rebuilding a turntable and it is extremely important to retrieve detail by connecting the tonearm as rigidly to the spindle of the platter.
So here we have the stators flopping around under transients because of the give in the tape. Anybody actually made a cell both ways to see if the rigidity can be heard on transients? As soon as I get another working panel and finally get both channels of my ML speaker working, I can listen to all my old material. The kind of stuff I am taking about are rim shots, wooden blocks, tight guitar, and any type of percussion. If the reproducer is outstanding, you can hear the air around each instrument. But if your system has a lot of mechanical hysteresis, and almost all the cone speakers I have heard do, you lose some of that air. This has nothing to do with room bounce, my headphones reproduce it and there is no room bounce there.
An example of the best reproduction of resolving transients are the chimes on the album Dark Side of the Moon. When i first heard it on an high fidelity system I thought this sounds exactly like a real grandfather clock like my grandfather had.
Not sure how much the relative non-rigidity of the spacers affects the sound in practice. I look forward to reading your appraisals. Plexiglass is indeed a favorite spacing material for DIY ESL builders. I suppose you could also look for stator movement during transients using high speed motion photography or other instrumentation, if available.

One of my favorite sources of transients galore is a recording by Murray Perahia and George Solti of Bartok's Sonata for Two Pianos and Percussion. I've heard the work in concert a few times, am familiar with the sounds of all the instruments, and have a nice grand piano in the next room (though not played by the likes of Perahia) but acoustic memory is unreliable. I think it sounds fine on my ML's though.
 
First, thanks very much for posting this thread; fascinating stuff.

An example of the best reproduction of resolving transients are the chimes on the album Dark Side of the Moon. When i first heard it on an high fidelity system I thought this sounds exactly like a real grandfather clock like my grandfather had.
The latest remix in Atmos is spectacular. I have the 5.1 SACD, and the Monoliths get the transients perfectly. The panels seem to have little resonance internally. But the mount to the frame is another thing, and at 400Hz to 700Hz, there are some resonances that I need to dampen with something I can fit between frame and panel.
 
One of my favorite sources of transients galore is a recording by Murray Perahia and George Solti of Bartok's Sonata for Two Pianos and Percussion. I've heard the work in concert a few times, am familiar with the sounds of all the instruments, and have a nice grand piano in the next room (though not played by the likes of Perahia) but acoustic memory is unreliable. I think it sounds fine on my ML's though.
Absolutely right. This one is a feast and a good test. I used it to validate the timings and integrate the MidBass Modules into the system.

I know it by heart, as my dad was a Julliard-trained concert pianist (but worked as a university professor in Costa Rica, so this was a side gig). He and a fellow pianist prepared the Bartok piece to play with the national symphony. In addition to hearing the one-paino side while he practiced on the Steinway grand at home, I went to several practice sessions at the concert hall and to two of the performances. And you are right. It's amazing live.

And I assure you, with a large ML ESL set in a well-treated room with plenty of clean low-end and strong mid-bass, this piece shines. I'm off to my HT to hear it now; thanks for the inspiration.

PS- This is the disc I'll be using, the 5.1 SACD of it: https://www.discogs.com/release/6438599-Bartók-Sonata-For-Two-Pianos-And-Percussion
 
Last edited:
Doesn't the Logitech Transporter have a USB in? Most DACs today are recognized in Windows or Linux (and probably MAC OS but I have no experience) as a sound card. These days, I run my computer's HDMI output into my Emotiva a/v proc to use REW. I also listen to analog sources, all the time, through a separate analog preamp, but using the analog output of a computer would generally not produce good results.

I looked at the Transporter when I was shopping for DACs years ago (having already amassed a fleet of Squeezebox players), but went with the Benchmark DAC1 Pre instead (which definitely can act as a sound card). The deal breaker for me was it didn't have a discrete remote command for inputs, only to toggle from one input to the next (am I right?) so programming an "activity" in a universal remote would have been kludgey or impossible.
The Transporter doesn't have a USB input but it has toslink, coax PDIF, S/PDIF & AES digital inputs, all of which go thru its DAC to an analog output (only). None of its digital inputs can be routed to its digital output.

In my setup, I use the Transporter as a digital preamp; feeding its digital output directly into a digital input on the DBX Driverack Venu 360 DSP/Crossover.

In this way (no analog preamp) one D/A conversion and one A /D conversion are eliminated, leaving a just one D/A conversion in the signal path (crossover > D/A > amps).

I suppose I could feed the REW into one of the digital inputs on the Transporter and use the Transporter's analog out, then into an analog preamp, then out to an analog input on the DBX unit, in which case there would be D/A + A/D + D/A conversions in the signal path.

That would work but it's not the signal path I use for music playback, so would the test results then be representative?

If I could figure out a way to stream REW to the Transporter's digital out, the results would then be representative of my music playback signal path. The Transporter has an ethernet connection so there may be a possibility there.

I don't know a way to connect REW into my single-conversion music playback signal path, and I don't want to revert back to multiple conversions in the signal path for music playback.
 
Last edited:
It was awesome!

And yes, found a resonance, but it was stuff in the room, not the gear 😀

IMG_5884.jpeg
 
I did a search and found this ideal coating material, very much like was used in DW speakers. I calculated that the charge diffusion time constant should be about 100ms. This would is roughly 5 times the period of 50 Hz. If you use the panel at a higher frequency you can chose a shorter time constant. Given that the stator to cell capacitance is 120pf, the resistivity needs to be about 850 Mohm per square.
So this liquid will fit the bill and you get a big quantity in each can. But it is even better! you need to dilute this 5 times to get the high resistivity and now you have 4L of liquid to share with all your friends. The instructions are clear, clean with isopropyl alcohol, this will not only clean, but slightly etch the membrane to make the solution stick for decades. I believe I got a sample of this years ago. The person on the phone told me that is what Beverege uses on their speakers. Remember those days when you actually called up someone and they helped you out? MG Chemicals 844AR ESD coating, 850mL.. No reason to go and find some fancy ESL builder that sells you a few ounces for a high price.

https://www.amazon.com/MG-Chemicals...f7644529fb7430a39e658c8d1fe70801&gad_source=1
Thanks for the tip. I had not known about that particular ESD coating.

I've been using Licron Crystal aerosol spray, which dries to about 2 microns thickness with E6-E9 resistance. So far it's worked fine for my hybrid ESLs.

Do you happen to know whether or not Beveredge dilutes the B&G solution in their application?

I read a post somewhere about someone having diluted Licron Crystal and it didn't work too well for them-- lost its conductivity after a year or so, as I recall.

I've always applied the Licron Crystal in one coat, without dilution and I haven't had any problems so far (I can vouch for five years, at least). It's advertised as a permanent ESD coating but I doubt the manufacturer tested it under the same conditions that it would see in an ESL.
 
The issue is, if you read articles on the web, that Mylar or its derivative have a low energy bond, meaning in my understanding, that they repel liquids on their surface. So it appears that even if the spray and other stuff is easy to apply, it won't last. The trick is to modify the surface to make it accept the coating. Of the two suitable liquids for home use to treat the surface is isopropyl alcohol or sodium hydroxide ( as in oven cleaner). Both etch the surface in preparation for the coating. I did not see how ML treats their surface, it's not shown in the video. The fact that their coating seem to evaporate more or less around ten years means that no surface treatment is used. The DW cells don't deteriorate because of surface coating going away, but because the high voltage creates pockets of corona almost all the time until there is a ground path to prevent the cell from biasing, or the area where the bias feeds the diaphragm attaches becomes so high in resistance.
I don't know what the actual ingredient that conducts in the various anti-static spray is. I looked at the coating from the seller in Australia, it is water based. I don't have any faith that it will last based on the chemical properties of the plastic. i read many years ago people used soap. Apparently this has a one year life. One thing I like about cell construction where the rivets are replaced with screws, you can take a cell apart unless its stuck in a frame surrounded by black corona dope, lol. The diaphragm on DW is only glued to one side of the cell frame. You separate the two frames and so you have access to it. After 40 years the surface resistivity is still intact. Mind you that the cell has not been exposed to any high voltage fields, but it shows that the coating chosen is sound (no pun intended).
My electrometer to measure surface conductivity.
 

Attachments

  • Electrometers.jpg
    Electrometers.jpg
    507.4 KB · Views: 0
So you'll be feeding into a distortion analyzer? Careful not to burn that out!

I've used a standard 1000:1 hv probe to measure the bias of my CLS II's. But since the impedance of the bias supply is so high, I don't consider the measurement reliable. Only for comparison shopping, when I wasn't convinced one of the bias supplies was turning on with a music signal.
A 1000:1 probe are typically 1000 Mohm. I saw somewhwere that the bias supply has a 33Mohm resistor in series? Not sure about that, but based on this resistance, you could just calculate a voltage divider.
In my case I wanted to measure the output of a 100:1 stepup transformer, and if you input 1V from your power amplifier, then the output is only 100V, easily measure with a 10:1 oscilloscope probe which usually has at least a 300V rating
 
Here is the front of the XG-10 Mk2. The tweeter mounted on the big cross-brace in the middle, with the element in front of the Mylar membrane. The tweeter housing was supposed to be sealed, but I suspect that the gas did leak out through it.
The reddish color on the frame is the remnants of the special glue that holds the Mylar to the metal frame. Mylar is notoriously hard to glue, that is why capacitor manufacturers have developed all these special glues. Some companies won't sell to individuals. So to get small samples, you have to have company sounding name and ask for a sample.
 

Attachments

  • _DSC2587a.JPG
    _DSC2587a.JPG
    1.1 MB · Views: 0
Back
Top