Thursday, October 4, 2018

DIY Electrostatic Loudspeakers


This post is some of the very first content I produced for the www in 1996, and then updated in 2006  It has been on my web site for all that time, and I'm trying to port everything from there over here so I can shut down the web site.  Forgive me if the formatting is a little awkward.  The stuff below has been grabbed from the web site, and a few updates added.

Intro


In the 10 years since I last updated my ESL pages, a lot has changed and I have learned a little more about some new materials and sources so its about time for an update.  The good news is it is now easier than ever to make your own ESLs.  Many more people have made their own speakers and created web pages to show how, so there is now more help than ever before available to the nascent speaker builder.

This page does not have specific plans for how to make a particular size or performance ESL.  Here I try to present you with enough information and ideas that you will be able to decide for yourself how to make speakers that meet your requirements.

Principles of ESL Operation


You can learn more than you need to know about how ESLs operate by looking at any college physics text book.  It all boils down to this- like charges repel each other, opposite charges attract each other.  If you place positive electric charges on two objects then bring the two near each other, they will experience forces that push them apart because of the interactions of the electric fields that surround them.  If one of the charges is "negative" and the other "positive", the two charged objects will experience a force that pushes them together.  The force involved depends on the charges and the distances involved.  An ESL uses the force developed by the electric field between charges to move a very thin plastic diaphragm back and forth to produce sound.

The behavior of electric charges is similar to magnets when brought near each other, however the forces developed by electric fields are usually smaller than those we experience with magnets.  You can play with electric charges by rubbing a balloon on your head on a dry day.  When you lift the balloon, your hair will stand up to try to touch the balloon because the balloon has accumulated one polarity of charge and your hair has accumulated the opposite polarity of charge.  This is called the triboelectric effect and occurs whenever two dissimilar materials are brought into contact with each other then moved apart. ESLs are simply a way to keep electric charges separated but close together so they can do the work of moving a plastic film to produce sound.  The usual construction technique is to place a tight plastic film (the diaphragm) with an electric charge between two charged conductors (called stators) on which the charges alternate.  At one instant, the diaphragm is driven towards one of the stators, then when the polarity of the charge on the conductors reverses, the diaphragm is driven toward the other conductor.  The result is sound.

The charges are applied to the diaphragm and the stators using some simple electronics.  A very low current, high voltage power supply is used to charge the diaphragm.  Electronic components called transformers are used to apply the alternating polarity charges to the stators.  I'll explain more about those, later.

When an electric charge is applied to a metal object, the charge spreads out and the entire surface of the metal rests at a specific voltage (a measure of the quantity of charge divided by the distance separating two opposite charges) depending on the amount of charge applied and the distance to the opposite charge (charges normally come in opposite pairs- you don't usually have a positive charge floating around without an equal negative charge somewhere nearby).  Metals as said to "conduct" electricity because electric charge can move around freely on them.

Plastics don't normally allow electric charge to move, so they are called "insulators".  An ESL uses a plastic film for the diaphragm because some plastics are very strong and light weight.  The diaphragm won't move if it doesn't have an electric charge, so the first problem in making an ESL is how to apply charge to the nonconducting diaphragm.

In between metals which conduct electricity and plastics that don't there are other materials that conduct electricity very poorly.  Those materials are said to have high resistance to the movement of electric charge.  In an ESL, the diaphragm is coated with one of these materials.  An electric charge is applied to the coating and over time, it spreads out over the entire surface of the diaphragm.  The very high resistance of the coating prevents the charge from moving around on the diaphragm surface very quickly.  The charge is applied with a high voltage power supply that can develop as much as 5000 Volts.  All that supply needs to do is supply charge to the diaphragm which is very high resistance, so the current required from the high voltage power supply is measured in microamps.  It is not dangerous unless you happen to be in an explosive atmosphere.  Remember the triboelectric effect?  That is what is responsible for the "static" electric shocks you experience when you walk across a carpeted floor with rubber soled shoes and touch a doorknob.  You can easily generate more than 5000 Volts by walking across a room on a dry day.

The speakers most people have experience with use magnetic fields to move a paper, plastic, or metal "cone" back and forth to produce sound.  The most common configuration has a coil of wire mounted near the center of the cone, concentric with a very strong permanent magnet.  Current produced by an amplifier flows through the coil and sets up a magnetic field that interacts with the field of the permanent magnet and produces force that moves the cone back and forth.  The cone is not perfectly stiff, and if it is large, it is relatively heavy.  These properties affect the frequencies that can be reproduced and the distortion.  A very large, massive cone cannot be moved fast enough to reproduce high frequencies.  When the coil is driven with a lot of current, the cone flexes a little resulting in distortion.  Low frequency sound reproduction requires movement of a lot of air, so the speaker must be made large, and the cone will be heavy.  High frequency sound production requires that the coil and cone be made very light weight.  This is why so many speakers have separate drivers for the low, middle, and high frequencies.

Most electrostatic speakers use a single, large diaphragm to produce sound.  Even though it is large, it is very light weight because it is made of very thin plastic.  The very light weight means it can reproduce high frequencies very easily.  It can't move back and forth as far as a magnetic woofer, but what it lacks in distance is made up for in surface area, so it can reproduce relatively low frequencies also.  Distortion is very low because unlike a magnetic driver in which the cone is driven only from the coil at the center of the larger cone, the entire surface of the diaphragm is driven by the electric field between the stators.  So ESLs are capable of reproducing a wide frequency range with very low distortion.

As wonderful as ESLs are, they have some limitations.  The large surface area means the speakers tend to dominate whatever room they are placed in.  They usually are not as sensitive as "normal" magnetic speakers so they won't play as loudly.  The frequency response of ESLs is a function of their size.  The larger you make them, the lower they can go, but it is almost impossible to get flat response down below 50-60 Hz because the front and rear of the speaker are out of phase.  At high frequencies the radiation pattern narrows to a tight beam.  This results in rising response- about 6 dB per octave, and a "sweet-spot" that is suitable for one listener only.

The response issues can be overcome by electronic equalization.  The only way to overcome the low frequency response limitation is to crossover to a magnetic driver in a box at some low frequency.  This will allow the speaker system to play loudly and remain very low in distortion.  If you aren't concerned about flat response below 50-60 Hz (most magnetic speakers can't go lower than that either), you can use ESLs with no crossovers at all.  They sound amazingly lifelike!

Materials Required


Here's a descriptive list of the materials and some sources for them.  You can literally use almost any plastic for the insulators, any film for the diaphragm, and any metallic thing that full of holes for the stators and your speakers will make sound.  If you want to experiment make small drivers using everyday materials you have on hand.  Once you have some experience handling the materials you can order the exact stuff you need to make that super high performance driver.
You'll need some tools, of course - saws, screw drivers, paint brushes, drills, soldering iron, and etc.  Nothing exotic.
A lot of the materials come in quantities much larger than you'll need to make a couple speakers.  You can minimize waste by talking some of your friends into making some speakers with you.  Then you can split the cost of some of the items and save some money.

1) Audio Transformers- one or two per speaker. 

There are two ways to go.  You can get audio transformers made for vacuum tube amplifiers or specifically for ESLs, or you can use low voltage toroidal power transformers.  The power transformers are MUCH cheaper - you can often find suitable transformers at surplus shops for $5-10.  Look for transformers with dual 115VAC windings and a low voltage winding such as 3-9V.  Alternatively, look for any 30-100VA rated toroid with 115V or 230V windings that is not potted (embedded in plastic resin) and add your own low voltage windings by simply wrapping a few turns of wire through the core.

Transformers made for audio use may ultimately provide better performance, but cost a lot more.  If you want to use audio transformers, look for tube amplifier output transformers for tubes such as 7189, 6BQ5,6GW8, 6V6.  They will have low impedance windings (4 Ohms) to connect to speakers (you will connect them to your amp) and high impedance windings (8-10K Ohms) to connect to the tubes (you will connect them to your speakers).  The transformers should be rated for around 30W or so at 30 Hz.  They will have to be made for use in push-pull amps (such transformers are center tapped) or you will need two transformers per speaker.  Tube amp output transformers are available from Antique Electronic Supply, and other sources. Expect to pay at least $50 each. 
Transformers made specifically for ESLs are also available but they tend to be VERY expensive.  Check suppliers like AmplimoPlitron, or Sowter.

In recent years, people have been using small toroidal power transformers to drive ESLs.  25-50W power transformers are cheap on the surplus market, and if the core isn't potted, you can easily add your own primary windings by wrapping a few turns of wire around the core.  Here are some suitable transformers.

Toroidal power transformer that can be used for audio step-up to drive ESLs.  Use two per speaker and connect the HV primary windings to the ESL.  Wrap a few turns of wire through the hole in the core and connect the ends to the amplifier output.


2) Plastic Film for Speaker Diaphragms- large enough to make the size of driver you want to build, plus at least 6" extra all the way around to allow for a mechanical stretcher.

If you just want to make sound, you can use almost any plastic film you can find, including Saran Wrap.  If you want to make hi-fi speakers, thin Mylar or other polyester (5-6 microns) is the way to go.  It can be obtained from companies that make plastics for industry- this film is commonly used to make capacitors (don't get metalized film!).  I bought a roll that is 1200 m long by 1 m wide for about $85 in Japan about 15 years ago.  It can be hard to find, but there are some hobbyist sources and you can even get it from McMaster-Carr.  If you buy stuff specifically for ESLs, expect to pay too much.  Buying stuff specifically for ESLs is like buying parts for Ferraris.  The seller knows you expect to be robbed so they try not to disappoint you.

3) High Resistance Coating for the Diaphragm.

I used to recommend things like powdered graphite to make the diaphragm slightly conductive.  It is still a good, cheap option if you are just trying to make a science fair project, but you cannot get very high resistance using graphite.  It will work, but at very low frequencies you can expect the speaker to distort a bit.  Powdered graphite for lubricating locks is available from your local hardware store for about $2 for enough to make about 50 speakers. Graphite has to be rubbed into the film using cotton balls.  It is a messy and tedious process.

If you want to make a hi-fi speaker that you will use for years, use Licron (or Licron Crystal).  It is a "permanent" spray-on anti-static coating that adheres to low surface energy plastics such as polyester.  It provides a much higher resistance than you can ever hope to achieve with graphite and it is 100X faster and easier to apply.  It costs about $40 for a 10 oz spray can but you'll be able to make more than 100 speakers with it.  It is NOT clear, and tends to make the diaphragm look cloudy, so if your diaphragms will be visible, Licron may not be the way to go.

I have seen others claim that diluting white glue about 1:5 and doping it with a few ions (antistatic fluid or some salts) makes a very high resistance diaphragm coating that sticks to polyester.  I have not tried it yet, but if it works it will be an ultra cheap way to coat the diaphragms.  They tell me it is practically invisible and weighs less than a Licron coating, too.

4) Perforated Aluminum or Steel for the Stators

You need one piece for the front and the back of each driver. It should be flat and have about 60% or more open area (holes). Hole size? The stuff I use has holes that are about 3 or 4 mm diameter. The "rules of thumb" say don't use holes larger than about 1/4". Check your local Yellow Pages phone book for listings under Perforators, or Sheet Metal.  Your local hardware store may have some available also.  Aluminum is much easier to cut than steel, and it is much lighter weight, but costs more than steel.  If you buy from a perforator you can get them to cut the metal to size and roll it flat for you.  You can order perforated sheet metal (and most of the other materials to make ESLs) from McMaster-Carr.

You can use aluminum window screen glued to a support frame such as a plastic fluorescent lamp grid.  Some people make stators by wrapping wire around a wood or plastic frame.  I don't think it provides better sound and it is a lot more work, but it looks different from perforated sheet metal and that is sometimes reason enough to go to the extra trouble.

5) Insulating Material for the Driver Frames- this is used to hold the two stators apart.  

The diaphragm will be sandwiched between and attached to the insulators.  For most ESLs, you need insulators about 2-4mm thick.

Use PVC, acrylic, fiberglass PCB stock, or almost any other insulating material you can get readily and cheaply.  Fiberglass is hard to cut (you need a carbide saw blade), and the dust from sawing is a health hazard, but epoxy will bond to it and the stators very well.  It will make a very stiff driver, and stiff is good. 

Acrylic or other plastics are easier to work with, but you may have some trouble gluing the stators to it (try contact cement). I have used both acrylics and PC board and for all it's trouble, I prefer the PC board material.  You can get fiberglass from a PC board company- try to raid their scrap pile- and get them to cut the pieces to size for you.  We'll talk about thickness later.

6) Adhesives - this is the trickiest part of making ESLs.

You will have to do some experimentation and research to select adhesives that work with the materials you use to make your speakers.  Epoxy will bond to metal stators and to fiberglass insulators, but not to other types of plastic, and forget about using it to attach the diaphragm to the insulators - it will not bond to polyester at all.

Contact cements are quite versatile and the right one can be used to bond almost everything in an ESL.  One that I have had a lot of success with is 3M  Scotchgrip 4693 or 4693H.  It is formulated to bond with low surface energy plastics such as polyester.  You put a little on one or both surfaces to be glued and let it stand for a few minutes. Then you put the two surfaces together and it bonds instantly.  The bond is so good that the film will tear long before the glue lets go.  The only disadvantage is that once you've assembled the driver using contact cement, it is hard to tear it apart and rebuild it, should the need arise.  You can get 4693H in 5 oz tubes from McMaster-Carr for about $8.  Or from Amazon.com.  One tube will be enough to make several large speakers.

7) High Voltage DC Bias Supply (1000-5000VDC, almost no current).  One for each speaker you build.

I recommend using a DC-DC converter that uses a low voltage input supplied by a wall-wart.  Such converters are usually very small and very safe because they can't supply much current.  They often provide output voltage that is proportional to input voltage so you can adjust the bias to suit your speakers.  See the bias supply design page for more details.  Several companies make the DC-DC converters or you can use a surplus power supply from a copy machine, though they are usually larger and not as flexible in terms of output voltage adjustability.  Small DC-DC converter modules come from companies like Emco High Voltage and Pico Electronics.  Check on-line electronic surplus dealers (Herbach & RademanAll ElectronicsAlltronicsC&H SalesElectronic Goldmine, and etc.) for copy machine supplies.

Now you can buy small, cheap HV DC-DC converters from China for about $5 each that will take in 6-15VDC and output 2000-6000 VDC.  A small wall-wart can be used for power input, and if you want variable output, add a few $ more and get a variable regulator.  More specifics on this, below.

8) Metal Tape - to connect the bias supply to the diaphragm.
Use about 2 feet of copper tape or other metal foil tape.  You can get it at McMaster-Carr.

9) Optional- insulating coating for the stators

If you use very high bias voltage to make your speakers sensitive, you will sometimes hear whining due to corona discharge around the sharp edges of the perforations in the stators.  This can be reduced by insulating the stators.  I've heard that latex house paint works fine, but be careful not to plug up the holes with it.

10) A Diaphragm Stretcher - build it yourself

You need a bicycle tire tube, a piece of plywood a little bigger than your drivers will be, some wood strips to screw to the perimeter of the plywood, and some double-sided adhesive tape.  About 18 years ago, when I made my first ESLs, I invented a very simple diaphragm stretcher.  It uses common materials and is very quick to build.  Best of all, it will put as much tension on the film as the film can handle and stretches the film more or less evenly.  It is so simple its a miracle that no one thought of it before me.  You simply stretch a bicycle tire tube around the perimeter of a table.  Next, you apply double sided adhesive tape to the underside of the table, then lay the diaphragm film out on the table and fold the edges under and attach them to the tape.  Finally, you inflate the tube.  It works every time. If your drivers have a perimeter greater than a tire tube, just stretch the tube until it fits.

You'll find more information about making the stretcher is on the Diaphragm Stretcher page.

This is an exploded view of an ESL.  Perforated stators are blue, insulators are green, metal tape is orange, and the diaphragm is red.

There are a bunch of different ways to stretch a plastic film tight and you will see a few of them if you look around at DIY ESL sites on the web.  I have tried the metal rods with screws at the corners and heat shrinking and my experience has led me to conclude that if you want to make speakers that are sensitive enough to use normal, 30-50W/channel power amps, the only way to get adequate tension on the diaphragm is the pneumatic stretcher that I invented about 25 years ago.  It is so easy, fast, inexpensive to make, and so effective, that once you try it you will wonder why you ever tried any other method of stretching the diaphragm.

The pneumatic stretcher consists of a table with a bicycle tire tube stretched around its periphery.  The table has a "lip" that borders it and the tube is stretched over that.  The inside of the lip has double sided adhesive tape stuck to it.  The tube's valve passes through a slot or hole in the lip that surrounds the table.  My stretcher has pretty sharp corners and edges.  I recommend that you round the corners and edges on your table when you make it.  Also, the lip on my table is just taller then the uninflated tube.  I recommend that you make the lip a bit taller- maybe 75mm (3 inches) or so.  Trace the outline of one of your stators on the table so that you'll know where to apply glue to the film.

Top view of the diaphragm stretcher table..  it's just a piece of wood with a rim and a bicycle tire tube stretched over it.

Bottom view of the stretcher table.  The green stuff is double-stick tape. The film goes on top of the table and is pulled to the underside and stuck to the tape.  Inflating the bicycle tire tube stretched around the table stretches the film tight.

The diaphragm is stretched by laying it on top of the stretcher table, smoothing it flat, then folding it over the tube and lip of the table and securing it to the tape that lines the inside of the lip.  Once the film has secured to the tape all around the lip, you attach a bicycle tire pump and inflate the tube.  It will stretch the film VERY tight.  You can stretch it so tight that it will break so be careful.  You want it to be almost, but not quite, tight enough to break.  You will have to experiment to see how much stretching your film can take.  Unfortunately, the pressure in the tube is not high enough to register on a normal bike tire pressure gauge (I am using a Blackburn TP-1 pump with attached gauge), so you can't easily use the air pressure as an indicator of the stretch.

Underside of the table with the film attached to the tape before applying tension.

Top side of the stretcher table with film in place but not yet stretched.

Once the film is stretched tight, there will be no wrinkles.  You apply some Scotch Grip 4693 to the film where the insulator will be attached, apply some to the insulator/stator assembly, wait a few minutes, then set the insulator/stator down on the film.  The two will bond instantly.  It takes 24 hours for the bond to achieve its ultimate strength, but after a few minutes it is strong enough that you can cut the diaphragm/insulator/stator assembly free.

The next step will be to coat the diaphragm with a high resistance coating.  I like Licron.  Just mask off the insulator to protect it from the Licron and spray it on.  Let it sit for an hour or so to dry and it will be safe to handle.  Peel off the masking tape, and complete the assembly by attaching the other half of the driver.

I recently performed some tests that demonstrate that the table provides uniform tension.  I drew a series of lines on the table with a green pen, then attached a piece of diaphragm film, inflated the tube just enough to take out most of the wrinkles, then traced over the lines on the film with a black pen.

Here's what the film looks like when you start to stretch it and just take out the wrinkles.  I marked black lines on the film that overlay the green lines on the table.
Here's what it looks like when you stretch the film tight.  Notice the tube is thicker in some places and narrower in others.  That helps keep the tension even all over the film.  Also notice that the black lines show how the film has stretched in all directions.
This is a close up of the upper right corner.  
Here's the lower right corner.
This is what happens if you over inflate the tube.

ESL Driver Construction

Warnings:
Before we go any further, I want to warn you about a couple things you may not be aware of.  Electrostatic loudspeakers use high voltages to operate.  They require bias of up to 5000 VDC and use AC voltages up to 5000 V.  The DC bias is sometimes supplied by a power supply running off 120VAC electrical circuits which can be dangerous.  The AC voltages used to drive the ESL are usually produced by connecting your stereo amp to an audio output transformer. The voltages produced by the audio transformer are dangerous!  Don't screw around!   If you have little kids in the house or if anyone might for any reason touch the speakers while they are operating, design your speakers so that it is not possible to come into contact with the drivers. If you don't know how to handle high voltage circuits, enlist the help of someone who does, or buy one of the commercially available ESLs.

Making the speakers

Building ESLs involves the use of tools and materials that if handled improperly can be hazardous.  Please make sure you know how to use these things before you begin.  By all means, use safety glasses at all times.  It would be foolish to trade your vision for the pursuit of audio ecstasy!

Step 1. Decide how big to make your drivers.

It is generally easier to make small drivers than to make big ones, but with small drivers you may need several of them so it takes more effort to mount and wire them. You need one insulator for the front and one for the rear of each driver. Ideally, the insulator frames should be cut from a single piece of insulating material. But they don't absolutely have to be made from a single piece. Be sure to plan and leave room for electrical connections (3 wires per driver) and mechanical mounting. I have built many drivers using different geometries and found that the following thicknesses and bias voltages will result in drivers that closely match the sensitivity of conventional boxed bass drivers without the addition of a lot of attenuation in the low frequency section of your crossover:

ESL use     total ESL area     DC bias     insulator thickness
mid/tweet     >2 sqft                 2kV             1/16"
full range      >4 sqft             3-5kV             1/8"

The insulator thickness to use is a function of many variables. If you want to reproduce low frequencies (down to 100 Hz or lower) you need to have room for the diaphragm to move. That means thick insulators. You will also need to use high bias voltage and high driving voltages (two transformers) to get reasonable sensitivity.

The mechanical force on the diaphragm varies as the square of the distance from the stator plates. That means that if you double the thickness of the insulators, you need to use four times the voltage for equivalent acoustic output. It isn't easy to make full range ESLs, and they almost never deliver enough bass. You need really huge surface areas to get bass, but that increases the capacitance of the driver and can limit high frequency response. You can improve the bass by using electronic equalization and mounting the drivers in the corners of a room. There is plenty of room for experimentation.

For midrange/tweeter drivers to be used in a hybrid system, there is considerable flexibility in the insulator thicknesses and spacing, bias voltage, and driving voltages. 1/16" PC board material is extremely common and low cost so it is almost ideal (except for the difficulty in cutting it) for this application. * 1/16" is easily enough room for the diaphragm to produce ear splitting volumes at frequencies down to 300 Hz or so, using a single transformer to drive each speaker.

Another benefit to using PC stock is that it is usually metalized on one or both sides, a feature that can be very useful when making electrical connections to the drivers. It will be best to have one insulator frame metalized on both sides, and the other metalized on one side, but we can make due with any material, even unmetallized.

There is a "rule of thumb" about the dimensions of an ESL that relate to the insulator thickness. The rule is that the diaphragm should be supported at least every 100X units, where X is the thickness of the insulator pieces. 'Supported' means that you should put insulating strips in the driver to support the diaphragm in at least one direction. 'One direction' means that long narrow drivers are OK. If you use insulators that are made from 1/16" PC board stock, the diaphragm should be supported every 4-6 inches. If you look at Martin-Logan ESLs you'll see they have support insulators every 4-6 inches and that they are unevenly spaced, presumably to move resonances of each section to different frequencies.

One way to make the ESLs showing the use of PC board stock.  Drawing not to scale. Electrical connections are soldered to the copper pads (yellow) along the bottom edge of the driver. Be sure to extend the board beyond the stators (blue) to leave room for hardware to mount the driver on some sort of frame.  It isn't necessary for the copper traces to loop around the perimeter of the driver.  A single contact point is really all that's necessary.
Step 2. Electrical connections

You will need to make an electrical connection to the diaphragm. This can be done in any number of ways, but remember that you must maintain a high voltage potential between the metal plates and the diaphragm.  You may want to clean the insulators with alcohol and a very clean cloth before proceeding.

The image above is the top sides of two pieces of PCB material used to make an ESL.  The one on the left only needs copper (yellow) on one side, the piece on the right needs copper on both sides.

The picture above is the bottom sides of the two PCB pieces used to make an ESL.

The electrical connection is made by physical contact between a metal strip and the conductive coated surface of the diaphragm. The metal strip may be the copper on a piece of PC board stock used for the insulator (very rugged and solderable), or it can be a piece of aluminum foil, or Radio Shack burglar alarm foil tape (both somewhat delicate and not solderable). Just remember that you have to be able to connect a wire from the HV bias supply to the metal. Also, epoxy is generally not electrically conductive (there are conductive epoxies available, but they are usually quite expensive), so don't completely cover the metal with epoxy.

Here is a tip to help insure long life for your ESLs. When you connect DC bias to the diaphragm, connect the minus side of the bias supply to the driver and the plus side to the center tap of the driver transformer. If you connect it the other way around, you'll find that over time the metal electrode that connects to the diaphragm will corrode like the plus battery contact in your car. 

Step 3.  Cut the perforated stators to the size that fits your insulators.  You have to use a shear or maybe a saw- be careful not to bend the material- you want it to be flat!  Once the stators are cut to size, epoxy them to the insulators making sure they are aligned with the electrical contacts.  Now your driver is in two pieces, two insulators with stators attached.

The image above is the two half-driver assemblies, ready to be glued to the diaphragm.  This is the outer side with the stators epoxied to the insulators.

The image above is the half-driver assemblies viewed from the diaphragm side.  The diaphragm electrical contact, yellow, will be pressed against the conductive coated side of the diaphragm.

Step 4. Stretch, coat, and attach the diaphragm to the insulators.

Stretching the diaphragm can be accomplished in two relatively easy ways. One way is to use a heat gun to shrink the diaphragm after it has been attached to the insulators. People have reported good results using this technique, but I have read that it "relaxes" over time and loses tension.

I use a stretcher table of the type shown previously. The table allows you to coat the diaphragm under full tension and allows you to make multiple drivers with nearly identical resonances (by inflating the tube to the same air pressure for each driver). To use it you lay the film on the table and use double sticky tape (green in the images below) to attach the edges of the film to the underside of the table.  You then pump a few strokes of air into the inner tube (blue in the images below) and watch as the wrinkles in the diaphragm disappear. You can put extreme amounts of tension on the film using this table, so be careful. Make sure you put a small hole through the table top surface to allow air trapped under the diaphragm to escape when you start pumping! 

How much tension is enough? That's a difficult question. The tension you use is a balancing act. It depends on the bias voltage you will use, the thickness and spacing of your insulators, and on the frequency range over which you intend to operate the driver. Usually you will want to operate the driver above its fundamental resonant frequency. If you want full range operation, that means you want the resonant frequency to be below 100 Hz or so. That requires low diaphragm tension but low diaphragm tension means you may have to use a reduced bias voltage or you may have the driver break into a low frequency oscillation where it pulls to one side, sticks until the diaphragm is discharged, then returns to the center until the diaphragm charges up again, etc., etc.

In reality the amount of tension you use isn't critical. Rectangular drivers have multiple resonances and you will always have some of them in your pass band. I have never been able to identify any of them by the sound of the driver when running test tones through it, and certainly never when listening to music. It may be possible in an an-echoic chamber or by using a FFT analysis of impulse response, but in your listening room there will always be room mode resonances and multipath effects that will dwarf the driver resonances. If the tension proves too low you can always reduce the bias voltage. 
OK, so you have the diaphragm under tension on the table. Now what? Cement the driver half without the diaphragm contact to the stretched film using contact cement (4693H if available).  Put some contact cement on the driver half, and put some on the diaphragm where the driver half will touch it.  Follow the instructions for the contact cement and allow it to dry for a few minutes, then put the half-driver down in the film.  Be careful- you only get one chance to get it right.  Pile some weight on it or clamp it and leave it sit to allow the cement to harden.  Once the cement is completely dry, you can release some of the air pressure from the tube, then cut the diaphragm around the edges of the driver to release it from the stretcher table.  
When I first made ESLs I tried to use powdered graphite to make the diaphragm electrically conductive.  It was messy and usually made a diaphragm with relatively low resistance- not good.  Then I discovered Licron antistatic spray.  It's a little bit neater and produces a very high resistance coating, and takes only seconds to apply.  Turn the driver assembly over and with the diaphragm side up, mask the driver mounting holes with some post-it notes.  Mask off anywhere else you don't want the conductive coating, too.  Then follow directions on the can and spray a light coating of Licron onto the diaphragm.  Let it dry.  Remove the masks, and place the other half of the assembly down on the diaphragm and screw the assembly together.  You could contact cement the other half of the driver to the diaphragm, but if there's ever a problem and you need to replace the diaphragm, you may have a difficult time separating the two halves of the driver.  If you screw the pieces together you'll be able to separate them again, peel off the old diaphragm and install a new one.


Step 5. Testing

Stand the driver up using Styrofoam blocks to insulate it or hang it from a frame using nylon cord. Connect the transformer(s) to the driver per the drawing below. Next, connect the bias supply wires to the transformer and the driver. Power on! If all is well you should hear a very quiet click or nothing at all. 

This is how you wire an ESL.  The 10 Meg resistor (or 20 - 100 Meg) connects to the speaker's diaphragm.  Diagram lifted from here.

You may hear a whining sound due to corona discharge which you may be able to locate by turning off the lights and looking closely at the driver.  Once your eyes have adjusted to the dark you may see faint blue sparks, probably coming from edges or pointed areas of metal. The cure is to reduce the bias voltage, or apply some insulating coating (finger nail polish works) to the point where the discharge is occurring. This problem can be avoided almost entirely by plastic coating the stator plates before assembling the drivers. If you coat them with plastic, make sure you leave some provision for making electrical connections to the metal.

The other thing you may see/hear is the diaphragm flapping back and forth because of insufficient tension. This can be cured in two ways. Lower the bias voltage or replace the diaphragm using higher tension. You may try using a heat gun to shrink the film more and put more tension on it before you rip it apart to replace it.

If the driver sits quietly, connect the output of your amplifier to the 4 or 8 Ohm taps on the transformer. At this point I cannot stress enough that you should never, ever, under any circumstances touch the driver while it is operating. You will receive a severe shock, and you will suffer burns from the tremendously high voltages produced by the transformer that drives the ESL. I have experienced this and can tell you that it hurts like hell (and burning skin stinks)! Don't do it!

Turn on the amp and play some music. Turn the volume up slowly. You should hear very low distortion music with little bass content coming from the driver. If not, turn the volume up. Sometimes the connection to the perforated aluminum is poor due to oxide on the surface of the aluminum. As you turn the volume up and the driving voltage gets high enough, it will arc through the oxide layer and suddenly you will hear the music very clearly. The newly "cleaned" connection will work virtually forever after this first "burn-in".

That's it. Wasn't that easy?

Consider this: The high voltage used to "energize" the speakers causes them to attract dust. When you're not using the speakers, you may want to turn off the bias supply to minimize this effect. You should also put a brush attachment on your vacuum cleaner and clean both surfaces of each speaker once in a while.  Quad ESL-63s used some of the diaphragm film material to make dust covers for the speakers.

Roger Sanders' article included an equalizer circuit to increase the low frequency output of the drivers. The circuit amounts to bass boost similar to what you can get by using the tone controls on your preamp. Sanders suggests that even in hybrid systems, the equalization is necessary to keep the speakers from sounding too weak on bass to lower mid-range frequencies. I have used the equalizers and operated without them and find that the sound without the equalizer is satisfactory. You may want to try using the drivers without equalization first, then add the equalizer if you think the bass/lower mid-range is weak.

If you want to make a hybrid system and you already have some speakers that provide reasonable bass, try using them with the ESLs before you blow big bucks or go to a lot of trouble making bass boxes. You may find that the speakers you have will work well enough, saving you a lot of money/time/effort. 

Support Frame Construction


Once you have drivers you're going to need something to hold them in place.  If you're skilled and equipped for wood-working you can design and build a piece of art that even your wife will approve of.  If you're like the rest of us, who have neither the skills nor the tools to do a good job with wood, what do you do?  You go to your local home improvement supply store or lumber yard and buy some 2" schedule 40 PVC pipe, fittings, and glue.  You can get the stuff to make two speaker frames for about $30.

PVC offers a lot of advantages.  It is cheap, easy to cut and glue, and it can be made to look OK by either painting or covering with cloth "socks".  The fittings are all made to accurate angles so it is really hard to screw up, and if you do, just get a few more fittings and some pipe and start over- it's cheap stuff!  2" PVC makes a sturdy frame that can be filled with sand, lead shot, concrete, or even mercury if you are so inclined (though I don't recommend it).  Sheet metal and wood screws will hold on to PVC quite securely.

All you will need to build and assemble a frame is a hand saw.  A rubber mallet or hammer is also handy to dry-fit parts before you glue them together.  Cut the pieces to the required sizes to fit your drivers.  5/8" of pipe goes into each fitting so start with your driver dimensions and add 10/8" to determine the final pipe lengths.  After cutting the pipes, use a rag soaked in acetone to wipe the red or black ink markings off the pipe before you glue the frame together.

Make the frame using 4 tees, 2 elbows, and 4 end caps as shown in the image below.  You can screw some wood strips or aluminum corner stock to the pipes to mount your drivers.  You can also attach dust cover to the frame if you have enough plastic film.  Dust covers are a good idea, and become a better idea as your speaker's bias voltage increases.

Support frame made using PVC pipe.  You can use wood, aluminum t-slot extrusion, or just about anything else.

PVC cement sets in seconds, so you need to be very sure that you can align the parts in a hurry.  Dry fit everything before gluing it to make sure your speakers will fit.  Use a hammer or rubber mallet to break the pieces apart again so you can glue them.

It would be equally easy, but a bit more expensive, to build the frame using aluminum t-slot extrusion.

Bias Power Supply


Many DIY ESL sites will show you schematics of voltage multiplier power supplies for the bias supply.  I even had some of those schematics on one of my pages, but I decided that such power supplies are a little too dangerous for most audio hobbyists.  A lot of the people making these speakers are not very experienced with electronics and don't realize the danger inherent in line powered circuits.  I can't tell you how many times I've been asked if it would be OK to use a neon sign transformer to make the bias supply.  The answer is NO!  DON'T DO IT!  Neon sign and microwave oven transformers are capable of supplying enough power to kill you.  I designed this supply to eliminate the danger.  It costs a little more to make than a supply made from scratch but it is far more versatile because the output voltage is variable.  It is very small- it was designed to fit inside the 2" PVC pipe frames that I was making for my speakers.

The heart of the circuit is the Emco G-series DC to HVDC converter module.  It is very small and fully encapsulated with pins for circuit board mounting.  I used a G-40 module which cost about $60 when I bought mine 10 years ago.  They are probably cheaper now.  This module is rated at 4000VDC out (at 250 uA) when 12VDC, 100 mA are applied to the input. Emco has modules in this series that go as high as 6000VDC out.

The output voltage is approximately proportional to the input voltage so by using an adjustable regulator circuit ahead of the module you can make a bias supply that can be varied from about 300V to 4000V.  Adjustable bias can be useful if you find that you didn’t quite put enough tension on the diaphragm of your speakers or if you have problems with corona discharge (I mean the whining sound that can sometimes come from the speakers, not excessive urination after drinking too many Mexican beers).

The adjustable regulator is made from inexpensive components including an LM317T 3 pin adjustable voltage regulator IC.  The complete schematic of the supply appears below, along with a PC board layout.  The dimensions of the board are 1.75" x 4".

You still need a HIGH VOLTAGE RESISTOR at the output of the supply to ensure constant charge (thus low distortion) operation of the speakers.  High voltage resistors are not the same as your standard, run of the mill carbon or metal film resistors.  Standard resistors will slowly self-destruct (open) if you apply too high a voltage to them (anything over 500V should be considered too high).  High voltage resistors are available from Caddock (try MX-431-20M-10%; go to www.caddock.com) and Micro Ohm (try RG-3H-U-20M-M; go to www.micro-ohm.com).  Surplus dealers occasionally have HV resistors.  Look for resistors that are 20-100 M Ohm.

Bias supply I used in the 90s.  The Emco modules are hard to get/expensive, but there are cheaper alternatives these days.

R1 is a 1% metal film part but feel free to substitute a lower tolerance carbon resistor. 590 Ohms is about what is needed for 12V maximum output. If you use a little bigger resistor, the maximum output voltage will go down slightly. The approximate maximum output voltage can be calculated from the following formula: Vout = 1.25 x (1 + R2 / R1).  When you wire the pot, be sure to short one end of the pot to the wiper contact. If the contact fails, you will still have 5k in the circuit and everything will still work.

You can apply any voltage from about +10 to +20V to the adjustable regulator and get some output from the bias supply. The 15V shown on the schematic is a nominal value and will allow you to swing the bias supply output voltage over a range of about 300 to the maximum output for the module you are using. The regulator won’t produce more than about 12V so it will not damage the DC to DC converter module no matter what voltage you apply to the regulator input.  If you turn the output voltage down low the regulator has to dissipate more power and it could get pretty hot so it's a good idea to put a small heat sink on it.

Update: you can now buy dirt cheap DC-DC converters via aliexpress and ebay that will do the job:

You can get these for about <$5 via ebay.

This is the Vin vs Vout for the module above.  Input current was under 15 mA at all voltages.

Variable DC regulator module that can be used to drive the HV converter module.  These cost $3. Use any 12V wall-wart you have laying around to power this module, or buy a new one for $5.



HV resistors available for $3 each here.


Bibliography

Links:
I was going to put a lot of links in here, but they tend to rot and I don't have time to keep up with all of them, so I'm going to suggest that you do a web search on your favorite search engine.  Look for terms like "DIY Electrostatic Speakers" and you'll get hundreds of hits.  Be sure to check out The Audio Circuit.  There's a lot of good information there.
One warning about audio web sites in general:  there's a lot of nonsense out there about audio and you're bound to run into some of it.  If you are new to electronics and/or audio, be skeptical about claims that just don't seem to make sense.  Most of the nonsense is promulgated by people who have a financial interest in convincing you that it is true.  It is safest to consider ALL information from such sites as suspect.  There are plenty of other, good sites, so don't waste your time on the nonsense unless you just want to have a good laugh.
The patent list below will be updated soon- check back again...

US Patents
You can search these and other patents at the US Patent Office

2,631,196 Janszen, March 1953
Description of an ESL that has on fixed electrode, and electrically divided diaphragm to control directivity, frequency response, and impedance of the speaker.

2,896,025 Janszen, July 1959
Describes and ESL manufacturing process using wire grid fixed electrodes, one on either side of the diaphragm.

3,008,014 Williamson and Walker, Nov. 1961
Describes a method of makig ESLs with low distortion and a method of separatng them into different sections to reproduce different frequency ranges.

3,014,098 Malme, Dec. 1961
Provides a good, detailed description of ESL operation, and a design that incorporates segmented stators to control directionality at high frequencies.

3,668,335 Beveridge, June 1972
Describes a servo controlled ESL mounted in an enclosure, to allow low frequency reproduction, and with an acoustic lens to control high frequency dispersion. Patent includes schematics of vacuum tube servo control amplifier.

3,668,336 Wright, June 1972
Describes an ESL with electrical connections on one side of the speaker only,

3,778,562 Wright, Dec. 1973
Describes an ESL mounted in a sealed enclosure filled with a gas that acts as an acoustic lens to control the directionality of the driver and allows higher
operating voltages than would be possible in air, thus increasing the sensitivity of the driver..

4,289,936 Civitello, Sept. 1981
Describes an interesting inverse ESL with one perforated stator plate and two diaphragms on either side of it. The drivers are triangular, with multiple units
assembled into a kind of geodesic dome structure for control of directivity of the speaker.

4,703,509 Kanchev, Oct. 1987
Describes an ESL that uses stators with resistive properties to control frequency response and directivity.

British Patents

1,234,767 Enock, June 1971
Describes electrostatic transducers in which the stators are insulated to allow the use of high operating voltages, thereby increasing the sensitivity of the transducers.

1,239,658 Bowers and Greenwood, July 1971
Describes an electrostatic speaker that is built used printed circuit techniques, in which the stators are insulated to allow the use of high operating voltages,
thereby increasing the sensitivity of the drivers.


Articles:

An Electrostatic Speaker System, Roger Sanders, Speaker Builder magazine, 2/80-4/80. A very good series of articles on DIY ESL construction.

Sound Radiation From Circular Stretched Membranes in Free Space, J. H. Streng, published in the Journal of the Audio Engineering Society, Vol 37 No.
3, March 1989. A detailed mathematical analysis of circular ESL driver resonances and frequency response.

Books:
(Click on the title to order from Amazon)

Audio Amateur Loudspeaker Projects, contains a good DIY ESL article that originally appeared in Audio Amateur magazine. Available from Old Colony
Sound Lab, PO Box 243, Peterborough, N.H. 03458. Phone: 603-924-6526, fax: 603-924-9467.

Electrostatic Loudspeaker Design and Construction, by Ronald Wagner, published by Tab Books. Describes operating theory and construction of ESLs.
Available from Old Colony Sound Lab, PO Box 243, Peterborough, N.H. 03458. Phone: 603-924-6526, fax: 603-924-9467.

The Electrostatic Loudspeaker Design Cookbook, by Roger Sanders, 1995. Available from Old Colony Sound Lab, PO Box 243, Peterborough, N.H.
03458. Phone: 603-924-6526, fax: 603-924-9467.

Loudspeaker and Headphone Handbook, edited by John Borwick. ISBN no. 0-240-51371-1. Contains a mathematically detailed description of ESL operation written by Peter Baxandall, including enough detail about the venerable Quad ESL63 to allow close duplication of the design (if you're so inclined).

11 comments:

  1. the $5 negative ion generators have two white output leads with what looks like carbon brushes on the ends.

    I'm assuming they are negative polarity so I would still need a + ouput lead to make a full circuit.

    Is there a way past that

    thanks Ronan Daly

    ReplyDelete
    Replies
    1. It's been a while since I tested those modules, but IRIC, one of the leads was positive and the other negative. Power the device up and if you get a spark between the two white leads, you'll know for sure. Figuring out which is the positive and which is negative is another matter... I used a HV probe on a multimeter so it was easy to tell. If you don't have a HV probe available, I'm not sure how you'd determine positive and negative, other than to hook them to your speakers and check the relative phase of the audio- if they are connected so the speakers are in phase, low frequencies will be there. If they are antiphase, there will be a distinct lack of low frequency output.

      Delete
    2. If you connect a neon bulb to the output (use a very high resistance), the pin inside the bulb that's connected to the negative side of the supply will glow.

      Delete
  2. Maybe I have not read correctly, I would like to know how much to pull the mylar, I am not clear on this point. Is there a way to vary the frequency response? Give it lower or higher? Thanks so much your tutorial is great !!! Great things.

    ReplyDelete
    Replies
    1. If the tension is not high, you will have to use a low bias voltage and that will make an insensitive speaker that won't play very loudly. So make the film tension as high as you can. That will raise the resonance frequencies but they seem not to matter much anyway.

      Delete
  3. Good morning.
    I find it very difficult to find mylar in europe.
    Would you have any references where you can buy it as a private individual?
    many thanks.

    ReplyDelete
    Replies
    1. Sorry, no, I don't know a source. Mylar is DuPont's brand name for polyester film. You might have more luck if you search for "polyester film" instead of "Mylar film". Years ago people used to sell the stuff in small quantities on ebay.

      Delete
  4. Thanks for this article. A while back my father gave me his pair of B&W DM70's, his since the 80's and broken for at least 15 years. The power supply I repaired several years ago, but I've always been a bit intimidated by the electrostatic units. The techniques here make it look much more approachable.

    ReplyDelete
    Replies
    1. The trick will be sourcing the polyester film. Good luck!

      Delete
  5. Modified Toroid transformer ??
    You suggest a way to diy the step up transformer:
    "Alternatively, look for any 30-100VA rated toroid with 115V or 230V windings that is not potted (embedded in plastic resin) and add your own low voltage windings by simply wrapping a few turns of wire through the core.
    This will act as the new primary winding."
    My question is: How will the amp respond to be nearly shorted on the output when the resistance is that low in the new primary winding ?

    ..Anders


    ReplyDelete
    Replies
    1. The amp responds to the impedance of the load, not just the resistance. A few turns of wire wrapped around a toroid core is a significant inductance but low resistance. The capacitance of the speaker on the other side of the transformer is also reflected back to the amplifier. The impedance is complex and you do need an amplifier that can drive it and remain stable. If your amp is speced to work with 2 Ohm loads, it will probably work, but even if it isn't, it may work fine anyway. That's a bit of an experiment. Adding a couple ohms of resistance in series with the amplifier can help stabilize the amp and make the whole thing easier to drive.

      Delete

Leave comments or a questions here and I'll try to post a response as soon as I can.