Anatomy of a high efficiency high output electrostatic loadspeaker.

[ESLFan]

I am starting this thread because this forum in general is host to many people that own electrostatic speakers, in addition to ML made ones. Some even build their own. I own a unique set of loudspeakers, that I had since 1978, and have been non-functional for about 25 years. Did not spend any time trying to rebuild them because of job and family priorities. The Dayton Wright loudspeakers probably went out of production in the early 80s. because of the design features and Mike Wright being a physicist they had serious reliability issues. They were basically a dream for an electrical engineer to keep repairing them., haha. I suspect that mike Wright has passed away since his web domain is up for sale and it seems all the design information he had posted is now gone. I was on the phone with him many times to get some idea how to repair them. I will share this information for those interested in building their own high output panels. This will be quite technical in nature.
First a few pictures of the cells themselves. For the completer speaker, there are many pictures on the internet, so I won't bother. I am going to show you the guts, so to speak .Below are the first of a series of pictures.
Item 1 is the bias feed. There is an additional resistor to prevent current hogging from the other cells. Because the resistance of the diaphragm is very high, maybe about a Gigaohm, the charge flows very slowly. Because of the current limit of the bias supply to keep arcing damage low, it can take up to a week to fully charge from completely discharged.
Item 2 are the stator supports and are recessed to prevent arcing. The dark spots on the stator are previous deposits of the diaphragm coating. You notice the curvature of the edges of the stator to gradually reduce the electrostatic field at the edges. The diaphragm has no coating where it curves.
Item 3 is the dead space to allow the diaphragm to flex as it is only attached at the top and bottom and NOT the sides.
The whole design relies on minimizing the parasitic capacitance. I have seen may designs that completely ignore this and attach the coated diaphragm to the edge where is adds useless capacitance but no sound. In this design, not even the stator is attached to the outside of the frame. Also the diaphragm is allowed to freely move the length of the edges for a distortion-less motion. The gap between the stator and diaphragm is about 0.25 inches. I believe that is larger than any other. The other thing to note is that there is no insulation on the stator. Mike Wright explained to me that if you add insulation which typically has a dielectric constant of between 3 and 5, it reduces the active gap and even worse it limits the dynamics and as the diaphragm moves closer to the stator, the electric field compresses in the dielectric and the motion becomes compressed at high excursion levels. I found this effect very obvious when listening to a lot of percussion and comparing it to other electrostatics. Some of these things don't work that well when using air, but Sulfur hexaflouride extinguishes the arc very quickly.

 

[ESLFan]

Here is a picture of the other side of the cells. You can see the curvature better.
1. Tabs for connecting bias voltage. There are two of them. The wire he used was bare so he relied on the gas to suppress corona. I would have still insulted them.
2. Damping strip on the stator that is also attached to the frame with an adhesive, silicon could also be used.as you can see where i had repaired one side.
3. is the connection from transformer for audio signal. Ech stator has their own wire.
The curvature of the stators can easily be seen. The diasphragm is stretched between the top and bottom by using heatshrink mylar that only shrinks in one direction. I have not looked for many years if this is still available. Dupont used to make this back then.

 

[ESLFan]

A picture of the diaphragm. It is 0.03mm in thickness. I had this lying around for about 20 years, so it is quite crinkled. As you can see the coating is a series of parallel tracks silk screened onto the Mylar. The side edges do not contain any tracks over the areas not directly over the stators. Mike told me to add the coating, you had to treat the Mylar to accept the coating. If I remember correctly, you can do that by briefly exposing the surface with a strong UV light, or use a chemical solution. Don't remember what chemicals he used. For the early prototypes, he used a bunch of Q-tips glued to a piece of wood to put down the stripes. Since these panels are designed to respond down to 30Hz, the charge time constant was chosen so that the electrical charge time constant was much longer than 0.3 secs. There are several tricks in this design to obtain low bass. One was to fill the cabinet with SF6. This also provided a longer acoustical path because the speed of sound travels much slower than in air. 436ft/s vs 1082ft/s. This makes the cabinet 2.5 times as big. Secondly, because the transformer coupling was really good, hence the huge size, at low frequencies the impedance reached about 300 Ohm. You then add a capacitor in series to resonate with the core inductance and set the frequency to the acoustic roll-off of the cabinet, but of course dampened the resonance with a resistor, and you get free bass boost without any extra power required from the amplifier. I believe it was set to about a 5 db boost. This method can of course be used with any electrostatic speaker provided you have enough excursion distance. This speaker used a 800uF bipolar capacitor, which some people said effected the sound. There also was a 3 uF film bypass capacitor in parallel with it.

 

[ESLFan]

A few more points to clear up before I go on further. Originally this was designed as a full range electrostatic speaker. But it ended up as a hybrid with a tweeter because one, amplifiers did not like driving a 4.7uF capacitor. To prevent amplifier destruction, a 3 ohm resistor was placed in series. Because the transformer is so big and wires size relatively large, the primary winding was really low resistance. Two, to keep the gas from leaking out, an outer Mylar film with a thickness of 10 mils was used which reduced the high frequency output further. So given that, the early versions had a piezo tweeter which coupled directly into the gas, then later a Panasonic leaf tweeter replaced it and was placed on the front surface. I never liked the sound of the piezo tweeter but the leaf tweeter sounded amazing. As a point of reference, the tweeter crossover point was 7kHz.
Because there were 10 cells in the last design, and they were mounted in a slightly spherical array, if you had rolled off the outer two panels on each side to produce just bass and the middle 6 for the rest, it would have made aplifier drive easier.
Also since each cell had its own current limited supply, if it arced, then the full bias volatge would still be on the all the others, so no reductioin volume was reaslly noticeble.

 

[Jazzman53]

Wow! ESL Fan, I really appreciate your posting details of the Dayton Wright ESL's construction.

I've always wondered about them and your posts are fascinating.

I heard somewhere that their transformers are really massive, about 40 lbs (?).

 

[ESLFan]

Going to take a picture of the inside of the box soon. The box with the two transformers and the HV power supply weight about 110 lbs. I carried one from basement where I used to have my sound room, upstairs almost 25 years. And I thought this is going to give me a hernia. I doubt than now I can move it by myself anymore. Oh but wait, the speakers themselves are 90 lbs or so each and two people need to grab them because of the bulk.
There is more descriptions and detail to come. I will point out some of the short comings, I always had a plan to improve upon them, but never got really motivated. The speakers are 1m x 1m by 6 inches deep, and really need to be placed 3 feet from back wall and 2 feet from sides. So you need a big room.

 

[Jazzman53]

One more question:
Can you even get the SF6 gas to restore the panels?

 

[ESLFan]

I will get a pic from the front, but they are heavy to turn around. With all the great design Mike put into it, he cheaped out on the support. To me the wooden frame to hold all those 10 cells is too flimsy, and when you knock on them they have a noticeable resonance. I think you can hear it a bit when you listen to them with music. As you can see they are arranged in a slight hemisphere to prevent too much beaming, but is ok, they are only producing up to 7 KHz. All the cells are glued with some black high voltage resistance black stuff. It is a huge effort to take out each cell. On top of that to repair the diaphragm, you need to drill out 9 or so rivets. Maybe one of you can tell me an easy to do that without melting the plastic.

 

[ESLFan]

Apparently in Georgia you still can get the gas. I checked earlier this year. Other states have stricter laws on greenhouse emissions. Each lbs of SF6 has 25,000 times the effect of O2. You have to order it ahead of time. It has been a while but i think I used to buy 10 lbs, each speaker takes something like 2.5 lbs and then I have something extra left over. It is really cool to fill them if all gas has leaked out. I make a small round hole on bottom, and another at the top. I use a small hose connected to the cylinder, turn on the gas and wait. Every so often I light a lighter and hold it against the top hole. If it goes out, the speaker is filled. While this is happening, the front of the Mylar holding the gas in will actually bulge out because the gas is actually heavy. You don't want to do it too fast so that the gas pushes out the air, sort of like filling the bathtub. To repair them I was lucky to get a 1000ft 36 inch wide Mylar HS65 as a sample. The salesman asked is 1000ft enough? LOL. This was in 94. Not sure if you can get samples that easily today.

 

[ESLFan]

If I were to rebuild them today, as I already started the process of ripping out the individual cells, I would use them just as bass panels and then make a narrow long strip either in center or at one edge, similar to the Apogee ribbon speaker design. Boy did those sound good. In 1996 they still cost $5000 and my wife said no.
As a side note, I have a second non functioning pair, so I will leave that one origional for the time being.

 

[Leporello]

One more question:
Can you even get the SF6 gas to restore the panels?

We use SF6 gas at the linear accelerator lab, where I work at a university physics department, all the time. It's not terribly expensive. I've ordered it myself.

 

[ESLFan]

Years ago I ordered it from Matheson Gas, located south of the Atlanta airport. I had to rent the cylinder, because I did not own one. One more piece of trivia, the thread on the valve is reversed. I threw away the adapter years ago, because i never planned to use it again. But something to know once you actually have a bottle and you want to connect it to something. .When I looked at this lately, the cylinders seem really expensive, say compared to the CO2 cylinders I buy for my beer brewing.
Try putting the gas into a balloon, it will certainly surprise people on how heavy it is.

 

[Russr]

@ESLFan - If Mr Wright did have info on the web, you can try using the "WayBack machine"... it has been crawling and archiving website info for decades!

https://help.archive.org/help/using-the-wayback-machine/

This is a fascinating thread... thanks for sharing

 

[Jazzman53]

I just sent my ESL builder/mad-scientist experimenter friend Jer a link to this thread.

Jer builds incredible stuff from practically nothing; winding his own trannies, and he posts his experiments on the DIY Audio Forum. He likes pushing the limits on voltages and SPL.

I recall one of his table-top size panels which used coated welding rod stators and a 13kV bias supply, and of course, he pushed it until it smoked, as usual

I would love to send Jer a bottle of SF6 and see what he does with it.

 

[ESLFan]

I found the schematic for the speaker finally. Hand drawn by Mike himself. Since the bias voltage is so high, to get maximum efficiency, it was up to the customer to reduce it if there was too much discharge, like a hissing sound. This of course is a bad move on his part. Most people cranked them up and left them The resultant discharge eventually starts to destroy the cells as the corona will carbonize the plastic making it conductive. of course then with reduced or low bias, you will loose output. One thing that impressed my friends when we listen to them in a windowless room, was to turn off the lights and watch the bluish glow of the corona discharge in many places. One problem was too that he decided to leave all the signal and bias wires bare relying on the gas to prevent corona. so at every junction where there was a sharp bend of the wire, a nice bluish corona discharge would take place. Now of course we know that under corona conditions the SF6 turns into many byproduct, some of them quite hazardous.

https://www.epa.gov/system/files/documents/2022-05/sf6_byproducts.pdf

Each tweeter has a level control so that you could balance the output to match what the cells were doing. So we can come to the conclusion that for most people it was set to taste and probably not accurate. There is a damping resistor across the capacitor to set the low frequency boost. Since the transformer has to drive the panels at low frequencies, it has three times more core than this:
file:///H:/Firefox/rk%201_68%2010cm%20imp%20251105.pdf
The maximum impedance is 3 times lower on the big transformer. If you are building just a midrange tweeter panel, then you don't need all that iron. If you happen to have an amp that has significant unbalanced dc output, the coupling cap helps and the 10 ohm resistor controls the current. Once you magnetize an audio transformer it is scrap because the mutual inductance will be so low that the primary windings can become a short for the amplifier.
One thing I did not mention was each cell is 61 pF capacitance. There are 10 so we have 610 pF total. This gets multiplied by the square of the 100:1 turns ratio, so now the amp sees 6.1 uF if it was not in series with a resistor. This is why so much effort was spend to reduce the parasitic capacitance to a minimum.

 

[Jazzman53]

Coincidentally;
This thread about repairing a pair of DW ESLs was recently posted on the DIY Audio forum:

https://www.diyaudio.com/community/threads/dayton-wright-esl.414329/#post-7719256

 

[ESLFan]

That is interesting. You may notice these cells are slighly different in design.
I got tired of adding gass all the time, so I supplemented the low output of the base with woofers. Unforntaly without gas the bias voltage has to be turned down and then 200W into 8Ohm or 350W into 4 ohm amplifier produces normal volume levels.

 

[ESLFan]

@Jazzman53. I am going to measure the primary inductance of the DW transformer and a few of my toroidal transformers i have lying around. On the Diy forum, I read that you have to use multiple power transformers in series to have a enough coupling at low frequencies, so you don't short out the amplifier. In the end a stepup audio transformer maybe the same cost as multiple ones if you build a full range ESL.
The chart above shows that maximum reactance is at 100 Hz and at 30Hz the impedance is still a respectable 90 ohms. So producing bass is extremly efficient and requires virtually no "real" power from amplifier. The phase reangle is about 60 deg.

 

[Jazzman53]

I should have thrown in a caveat about the DW's relative efficiency, considering their 1/4" d/s. From what I've read, they had some heft at frequencies lower than most ESLs can muster.

I do pretty well with the mechanical aspects of building ESLs but I have no background in electronics and I struggle with that part.

From what I gather, the amount of transformer iron needed to play low without reaching saturation increases exponentially as frequency gets lower.

The consensus among my ESL builder friends on the DIY Audio forum is that 100VA and 75:1 ratio are sufficient above 200Hz.

I use a tandem pair of 50VA 230V/2x6V power toroids with 6V windings in parallel on the input side and 230V windings in series as the secondaries.

I cross them over to the woofers at 265Hz with either a 24db (minimum) or a 48db filter slope. I push them pretty hard sometimes and I've never had a problem.

They sound much better on the top end than the expensive 100:1 IE core audio transformers I used previously.

If you're ever in Savannah, I invite you to stop in for a listen.

Must I tempt you with ESL **** ?

 

[ESLFan]

I am impressed. My daughter lives in Charleston and if time permits, next time I am there I will detour to Savanah on the way home.