Making a highly efficent ESL for play from 1khz-25khz

Reading up on ESL's ... it seems like a simple design that I've already enjoyed the sound from (ML clarities) so I would like to explore simply what is possible with them...

seems from my research that curved diaphragms are ideal for midrange/tweet duty while flat are good for bass extension..

obviously based upon this I would want curved

basically I'm looking for something that can cover 1khz-25khz with ESL's noted low distortion...

This is where I'm currently at:

I understand the construction of a ESL, but I don't understand how to make it more efficent.

the best ESL's I've come across are 90 db efficent... quite lack luster...

the only thing I know is to gain efficency with this design is to make the insulators thinner.... maybe 1/32"

I have 5 15" lambdas cover the missing range...

I'm also assuming that using a thinner mylar would increase efficency... but I'm unsure about this...

I just need it to be very efficent to match the SQ beasts I've already picked out... which are really efficent...

thanks
 
You can increase sensitivity of the speaker by either narrowing the air gap between the stators, increasing the bias voltage, or increasing the drive voltage by using a higher turns ratio transformer.

Here are the problems:

a) Decreasing the gap reduces maximum excursion and limits max SPL- shouldn't be a problem from 1 kHz to 25 kHz unless you use a ridiculously small gap.

b) Small gaps and curved panels don't go together. It is nearly impossible to put tension on the film in one direction without causing it to bow inward in the other direction.

c) Small gaps and high bias voltage don't go together. You need a lot of tension on the diaphragm to keep it from sticking to the stators. See comment b about high diaphragm tension. You can also run into corona discharge and arcing. Air can only withstand so much voltage before it breaks down and becomes a conductor.

d) Increasing the turns ratio in the transformer tends to increase inductance which leads to frequency response problems moving down into the audio band. You can only go so far with this.

There is one more option- use a higher power amplifier. No, the speaker doesn't need higher power to operate, but a higher power amp will be capable of wider voltage swings at the output.

I suggest you try making a flat panel driver (it can be small) to get some idea of what works and what doesn't. The experience will quickly familiarize you with all the variables. Then if you want to make something bigger you can do it.

Curved ESLs such as the ML units don't disperse high frequencies much better than flat panels. ML curves the panels to make them mechanically rigid. A flat, steel panel is very "unrigid". Curving it makes it much more stable. Try it with a piece of cardboard.

As long as the panel width is much greater than the wavelength of sound being produced, the driver will beam. It is simple "newtonian" physics - not even the quantum type.

Reasonable sized ESLs are entirely capable of reproducing sound down to 200 Hz or so. Why put a crossover at 1 kHz and split almost every human voice and musical instrument across multiple drivers? I don't think 15" drivers will go to 1 kHz...

Unfortunately ESLs are not the solution for high volume applications. They are about low distortion and wide response. If you want loud, make some horns.

I_F
 
I_Forgot said:
You can increase sensitivity of the speaker by either narrowing the air gap between the stators, increasing the bias voltage, or increasing the drive voltage by using a higher turns ratio transformer.

Here are the problems:

a) Decreasing the gap reduces maximum excursion and limits max SPL- shouldn't be a problem from 1 kHz to 25 kHz unless you use a ridiculously small gap.

b) Small gaps and curved panels don't go together. It is nearly impossible to put tension on the film in one direction without causing it to bow inward in the other direction.

c) Small gaps and high bias voltage don't go together. You need a lot of tension on the diaphragm to keep it from sticking to the stators. See comment b about high diaphragm tension. You can also run into corona discharge and arcing. Air can only withstand so much voltage before it breaks down and becomes a conductor.

d) Increasing the turns ratio in the transformer tends to increase inductance which leads to frequency response problems moving down into the audio band. You can only go so far with this.

There is one more option- use a higher power amplifier. No, the speaker doesn't need higher power to operate, but a higher power amp will be capable of wider voltage swings at the output.

I suggest you try making a flat panel driver (it can be small) to get some idea of what works and what doesn't. The experience will quickly familiarize you with all the variables. Then if you want to make something bigger you can do it.

Curved ESLs such as the ML units don't disperse high frequencies much better than flat panels. ML curves the panels to make them mechanically rigid. A flat, steel panel is very "unrigid". Curving it makes it much more stable. Try it with a piece of cardboard.

As long as the panel width is much greater than the wavelength of sound being produced, the driver will beam. It is simple "newtonian" physics - not even the quantum type.

Reasonable sized ESLs are entirely capable of reproducing sound down to 200 Hz or so. Why put a crossover at 1 kHz and split almost every human voice and musical instrument across multiple drivers? I don't think 15" drivers will go to 1 kHz...

Unfortunately ESLs are not the solution for high volume applications. They are about low distortion and wide response. If you want loud, make some horns.

I_F


I know they're not the solution for high volume apps... but can they be used in them?

it's either that or a very very large and power hungry array of speakers and linesource ribbons...

the 15" drivers I'm using are flat to 1 khz ;) they are in fact renowned for doing so...

what do you mean by the panel will beam??? is taht a good thing???
so wider panels are better?
wider than a wavelength at 1khz would be 1 foot... so would a 18 inch wide panel do well "beaming" 1khz-20khz?

what kind of power handling (generally) would a speaker like this have?
 
solid stator plates for high efficiency

How about building some ESL panels sideways with 100% solid flat insulated stator plates? They will require that one edge on either side of the diaphragm be left open so the air can escape when compressed by the diaphragm (front side/back side). You could run multiple sets side by side to boost output. You could even use this idea to make an electrostatic compression driver. Any thoughts??? Regards Moray James.
 
Re: solid stator plates for high efficiency

moray james said:
How about building some ESL panels sideways with 100% solid flat insulated stator plates? They will require that one edge on either side of the diaphragm be left open so the air can escape when compressed by the diaphragm (front side/back side). You could run multiple sets side by side to boost output. You could even use this idea to make an electrostatic compression driver. Any thoughts??? Regards Moray James.

it's an idea...

I wouldn't want to horn load it though as the SQ would drastically suffer....

running multiple sets is an idea... but that would require some sort of waveguide at least... though avoiding a horn is still ideal...

I'm scared about what this would sound like
:bigeyes:

it could either sound fine and be ungodly loud... or it could sound absolutely terrible
 
make a stack of pancakes!

Don't see why you would need to have a lens. Just parallel stack as many units as you want to get a composite panel of about one to two inches wide. If you made a single panel 4'X36" then you would have a lot of panel area there. Construction would be a snap. My concern wolud be with turbulance as the air exits and enters the panel through the small thin long gap along the side of the diaphragm. Regards Moray James.
 
Re: make a stack of pancakes!

moray james said:
Don't see why you would need to have a lens. Just parallel stack as many units as you want to get a composite panel of about one to two inches wide. If you made a single panel 4'X36" then you would have a lot of panel area there. Construction would be a snap. My concern wolud be with turbulance as the air exits and enters the panel through the small thin long gap along the side of the diaphragm. Regards Moray James.


yes... a simple half round wood piece could help with some turbulence...

it seems like that would be really sketchy to get right...hmmmm
 
Audiophilenoob said:
I know they're not the solution for high volume apps... but can they be used in them?

Sure, if you don't mind them not playing loudly, but that sort of goes against the idea of a high volume application, doesn't it?


what do you mean by the panel will beam??? is taht a good thing???
so wider panels are better?
wider than a wavelength at 1khz would be 1 foot... so would a 18 inch wide panel do well "beaming" 1khz-20khz?[/B]


Larger panels will play louder than smaller panels, but then the listening position will be very narrow (move your head an inch or two and you will hear the frequency response change). Wavelength at 20 kHz is a fraction of an inch. That means if the panel is wider than a fraction of an inch, it will be directional at 20 kHz. Practical panels, especially one's that you want to play at some lower frequencies, are going to be at least 1 foot wide. That means that the on-axis frequency response will rise with frequency above a few hundred Hz. Off axis (by just a few degrees) the highs will all but disappear.


what kind of power handling (generally) would a speaker like this have? [/B]


That's a function of the transformer used, and the stator spacing. You'll know you've hit the limit when the speaker arcs. I use 15W rated transformers and drive speakers with a 200W per channel amp and I can't make them arc unless there is a lot of low frequency content in the music. That all sounds impressive, but a dynamic or horn driver with that much power would put you in the hospital with ruptured eardrums.

These speakers don't require much power, they need voltage. They are very good at converting the input power to sound. There isn't much resistive material except the windings in the transformer, so the power going in mainly gets converted to sound. However, the the voltage and current of the applied signal are out of phase because the speaker consists mainly of reactive elements, so unless you swing a LOT of voltage, there won't be much power. That is why people often use high powered amp with ESLs.

I_F
 
I_Forgot said:


Sure, if you don't mind them not playing loudly, but that sort of goes against the idea of a high volume application, doesn't it?



Larger panels will play louder than smaller panels, but then the listening position will be very narrow (move your head an inch or two and you will hear the frequency response change). Wavelength at 20 kHz is a fraction of an inch. That means if the panel is wider than a fraction of an inch, it will be directional at 20 kHz. Practical panels, especially one's that you want to play at some lower frequencies, are going to be at least 1 foot wide. That means that the on-axis frequency response will rise with frequency above a few hundred Hz. Off axis (by just a few degrees) the highs will all but disappear.



That's a function of the transformer used, and the stator spacing. You'll know you've hit the limit when the speaker arcs. I use 15W rated transformers and drive speakers with a 200W per channel amp and I can't make them arc unless there is a lot of low frequency content in the music. That all sounds impressive, but a dynamic or horn driver with that much power would put you in the hospital with ruptured eardrums.

These speakers don't require much power, they need voltage. They are very good at converting the input power to sound. There isn't much resistive material except the windings in the transformer, so the power going in mainly gets converted to sound. However, the the voltage and current of the applied signal are out of phase because the speaker consists mainly of reactive elements, so unless you swing a LOT of voltage, there won't be much power. That is why people often use high powered amp with ESLs.

I_F

yes I've seen this is FR plots from an ESL... after around 6khz it slopes very quickly down... it's not exactly great

Seems like perhaps a wide ribbon tweeter would be more ideal than messing with ESLs
:xeye:
 
Is it really so?

Hi,

I´d like to give some comments on I Forgot´s thoughts.

"a) Decreasing the gap reduces maximum excursion and limits max SPL- shouldn't be a problem from 1 kHz to 25 kHz unless you use a ridiculously small gap."

Decreasing the gap rises SPL and even mostly max SPL (when driven with frequencies where no large amplitude will ocur)! The reason for this is the quadratically rising efficiency with decreasing gap, while the SPL max sinks linearly with amplitude.
The lower polarizing voltage with small gaps has no effect as long as field density (field strength) is kept the same, e.g 1.5kV/mm.


"b) Small gaps and curved panels don't go together. It is nearly impossible to put tension on the film in one direction without causing it to bow inward in the other direction."

They do! In fact, it doesn´t make sense to use wide gaps with curved panels because the curving will introduce distortion at high amplitudes. It is possible and useful to put high tension on the membrane. ML for example uses extreme tension on their panels (250Hz Fs on Prodigy panel!). But You have to support the membrane horizontally according the rules for dynamic stability. Built this way You can run such a panel even without HP-filtering to astonishingly high volume levels without the membrane hitting the stators! Driven over a HP-filter You can easily produce PA-levels with lowest distortion.


"c) Small gaps and high bias voltage don't go together."

Right and wrong. Less important is the voltage itself. It´s the voltage gradient that counts. A smaller gap will arc at lower voltages, but it is generally much more useful to lower all voltages (because of increased efficiency of the panel) than to increase them. All in all a efficient ´low voltage design´ works safer and sounds more dynamic. It decreases the need for very high voltage insulation for the Tranny (Bandwwidth) and the extreme stability and power on the amplifier side.

"d) Increasing the turns ratio in the transformer tends to increase inductance which leads to frequency response problems moving down into the audio band. You can only go so far with this."

Totally right! A high efficiency small gap design needs much less transformer ratio to sing, increasing bandwidth and sound quality and reducing the need for extreme quality extreme pricy Trannies.
You might even use cheap standard off-te-shelf toroids with excellent results instead of expensive handcrafted pieces.


"If you want loud, make some horns."

Undoubtly Horns can play loud. But a panel of ML Prodigy size won´t play much less loud but with probably less distortion and much greater bandwidth and less size and weight.

Because of the radiation pattern of slim and high panels (dipolar cylindrical) the sound pressure level over distance is more constant, decreasing with app. 3dB per doubling of distance, while conventional speakers loose 6dB. That reduces the problem of lower sensitivity (measured standardly on 1m distance). Measured on 4m or more a ESL panel can compete with a Horn!

To increase efficiency even more one could think of an ESL that uses a folded membrane quite similar to ESS´s AMT, thereby reducing the width of the ESL.

Yeah
Calvin
 
Moray, I would be concerned about your idea working if the gap was larger than a 1/4 wave at the highest freq of interest. You'd have to actually produce compression for the thing to work... if there was anything but a planar compression wave produced, I expect that you'd have constructive and destructive effects that would muck up the response. And, you have to "do something" with the sound coming out of the end with respect to that wiggly diaphragm... that junction would be non-trivial to make free of interference and diffraction effects.

And, on other posts:

As far as horns being efficient, they are. And you can make horns sound very good. But subjectively they are different from an ESL no matter what. I have and use both.

Now, there is a nice article in the JAES Speaker Anthology Vol I that shows the construction technique used for a horn loaded ESL! :D

That should provide the desired outcome? I'd chose a modern horn expansion over the older exponential shown in the design, of course.

As far as problems increasing the turns ratio on a transformer, it seems to me that it's not that hard to wind a small transformer for perfectly acceptable response with a rather higher than usual turns ratio... just more interleaving and better technique will get you most of the way, that and maybe a nice amorphous C core and ur home free... send me the samples for test and approval! ;)

The real question being asked is "effective sensitivity" not "efficiency." The efficiency of an ESL has been said to be very high - but that is a different matter than max SPL or sensitivity (power in for SPL out).

One obvious method to consider for an HF array is direct drive, since the ESL only wants voltage presented that's a nice way to go.

Now, what 15" driver is supposed to be flat out to 1kHz??

A) I don't believe it.
B) If it is "flat" on axis, it isn't anything close off axis
unless C) It has a "whizzer cone" in which case we have a whole 'nother can of worms, and D) if it is flat on axis out to 1kHz. then it has squat for bass response since you can't get a heavy cone to work well that high, it has to be a light cone a la old Altecs, Tannoys, or similarly constructed drivers.

Department of no free lunch.

Just my opinions though.

_-_-bear :Pawprint:
 
Oh!... regarding distortion and max SPL.

A really high quality horn system will trounce any ESL of normal size in both departments, assuming a very high quality compression driver for the horn and a modern horn (or maybe not even).

My compression driver & horn combo max's out at ~128dB @ 40 watts.

Even reading JBLs figures show that the 1% distortion isn't reached until fairly close to this SPL figure. The lower you run the compression driver, the more the SPL drops. (you lose the throat distortion component in addition to the normal reduction in distortion all dynamic drivers get from running *less*).

I don't think any commercial ESLs can even reach 110dB. If they can, they're really max'd out and that won't be low in distortion.

I'd bet the farm that at ~90dB that *my* horns at 1kHz will be lower in measured distortion than any ESL.

But that doesn't mean that the horns will sound like an ESL.

I like ESLs fine, I use them myself.

Different animals. Apply them differently. It's like a Porshe engine, useless in a Mack Truck, too big for a motorcycle. Perfectly wonderful for the proper application. Which is the point, use the proper tool for the job. There's nothing to gain in arguing about which is "best." (imho)

_-_-bear :Pawprint:
 
bear said:
Moray, I would be concerned about your idea working if the gap was larger than a 1/4 wave at the highest freq of interest. You'd have to actually produce compression for the thing to work... if there was anything but a planar compression wave produced, I expect that you'd have constructive and destructive effects that would muck up the response. And, you have to "do something" with the sound coming out of the end with respect to that wiggly diaphragm... that junction would be non-trivial to make free of interference and diffraction effects.

And, on other posts:

As far as horns being efficient, they are. And you can make horns sound very good. But subjectively they are different from an ESL no matter what. I have and use both.

Now, there is a nice article in the JAES Speaker Anthology Vol I that shows the construction technique used for a horn loaded ESL! :D

That should provide the desired outcome? I'd chose a modern horn expansion over the older exponential shown in the design, of course.

As far as problems increasing the turns ratio on a transformer, it seems to me that it's not that hard to wind a small transformer for perfectly acceptable response with a rather higher than usual turns ratio... just more interleaving and better technique will get you most of the way, that and maybe a nice amorphous C core and ur home free... send me the samples for test and approval! ;)

The real question being asked is "effective sensitivity" not "efficiency." The efficiency of an ESL has been said to be very high - but that is a different matter than max SPL or sensitivity (power in for SPL out).

One obvious method to consider for an HF array is direct drive, since the ESL only wants voltage presented that's a nice way to go.

Now, what 15" driver is supposed to be flat out to 1kHz??

A) I don't believe it.
B) If it is "flat" on axis, it isn't anything close off axis
unless C) It has a "whizzer cone" in which case we have a whole 'nother can of worms, and D) if it is flat on axis out to 1kHz. then it has squat for bass response since you can't get a heavy cone to work well that high, it has to be a light cone a la old Altecs, Tannoys, or similarly constructed drivers.

Department of no free lunch.

Just my opinions though.

_-_-bear :Pawprint:


the 15" driver in question is the lambda with the extended phase plug... people have played them with nice results to 2khz... 1khz is quite easy for them... and what they were made to do... inductance is practically non-existant in these drivers
 
bear said:
Oh!... regarding distortion and max SPL.

A really high quality horn system will trounce any ESL of normal size in both departments, assuming a very high quality compression driver for the horn and a modern horn (or maybe not even).

My compression driver & horn combo max's out at ~128dB @ 40 watts.

Even reading JBLs figures show that the 1% distortion isn't reached until fairly close to this SPL figure. The lower you run the compression driver, the more the SPL drops. (you lose the throat distortion component in addition to the normal reduction in distortion all dynamic drivers get from running *less*).

I don't think any commercial ESLs can even reach 110dB. If they can, they're really max'd out and that won't be low in distortion.

I'd bet the farm that at ~90dB that *my* horns at 1kHz will be lower in measured distortion than any ESL.

But that doesn't mean that the horns will sound like an ESL.

I like ESLs fine, I use them myself.

Different animals. Apply them differently. It's like a Porshe engine, useless in a Mack Truck, too big for a motorcycle. Perfectly wonderful for the proper application. Which is the point, use the proper tool for the job. There's nothing to gain in arguing about which is "best." (imho)

_-_-bear :Pawprint:


I've had my fun with horns in the past... I can imagine that there are plenty of horns that sound great...

a long ribbon line could easily be 106-109 db efficent.... but I dunno if I like the sound of them playing at 1khz...

Calvin had some great ideas and helped me come up with a plan on my test ESL
 
Yeah, the Lamda line was nice.

I'd still be wary of using a 15" higher than ~300 Hz. myself.

I'd suggest you do some tests on the driver's response anyhow.

My experience with "phase plugs" on drivers is that they have some benefits and some deficits. As things now stand, my view is that the deficits outweight the benefits, especially the higher up in freq they are intended to be useful.

Ymmv.

_-_-bear :Pawprint: