by
Mark Rehorst (markATrehorst.com) (replace AT with @ to email)
current version date: 6/7/96
previous version date: 1/15/96
Introduction
Electrostatic speakers are the just about the lowest distortion drivers that can be
made. But you already know about their wonderful attributes or you wouldn't
be interested in making them, so I'll dispense with the BS. I present here a
simple process for making ESLs. I have not included anything about crossovers
or cabinets. This is strictly a "how to make the drivers" article.
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 need
a DC bias of up to 5000 V and use AC voltages up to 5000 V. The DC bias is
usually 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 a vacuum tube amplifier type output
transformer. The voltages produced by the 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.
Section I 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. If would be foolish to
trade your vision for the pursuit of audio ecstasy!
What you need:
1) Transformers, one or two per speaker- use tube amp output transformers, 4
ohm:8K -20K ohm. I have used Tango CRD-8 ( 4:8KCT) transformers that I
bought in Japan. You can use transformers by Triad, Stancor, etc. Just get units
that are good for about 15-20 W at 30 Hz and give a large impedance
(i.e. voltage) transformation. Expect to pay about $50 each for transformers.
Tube amp output transformers are available from Antique Electronic Supply,
602-820-5411, and other sources.
2) Plastic film for speaker diaphragms- Mylar or other polyester, thin (5-6
microns), and large enough to make the size of driver you want to build. This
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 a few years ago. I have
used about 15 m of it so far. I have heard of people using Saran-wrap, but I
have never heard a driver built using it. If you're making small drivers, or
experimenting, try it! It certainly won't cost much...
3) Powdered graphite, dish soap, or antistatic solution to coat diaphragm.
Powdered graphite is available from K-mart or your local hardware store for
lubricating locks. It will cost no more than $2 for enough to make about 50
speakers. Graphite has to be rubbed into the film using cotton balls. Dish
detergent and antistatic solution will work also, and are easier to apply, but may
not be "permanent". I use graphite. Someone in Australia suggested that
drafting ink formulated for drawing on "film" (the draftsman's name for
polyester) will make a good, easy to apply, high resitvity diaphragm coating. I
haven't tried it yet, but applying a colored liquid ought to be easy and make it
easy to verify that it only went where you wanted it.
4) Perforated aluminum or steel
You need a piece for the front and the back of the 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 may cost a little 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.
5) Acrylic or fiberglass PC board stock for driver frame.
Fiberglass is hard to cut (you need a carbide blade), and the dust from sawing is
a health hazard, but epoxy will bond to it. Acrylic or other plastics are easier to
work with, but epoxy may not form much of a bond to them (contact cement
will probably work just fine). 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) Glue
Previously I recommended epoxy to hold the ESL together. Epoxy works fine
for attaching the perforated metal to the insulator frame. The problem with
epoxy is that it doesn't really bond to the mylar film. A little mechanical stress
can break the very weak bond and allow the film to peel away. This can be an
advantage. If you find that a driver doesn't work, if you assembled it with
epoxy it will be easy to rip apart and rebuild.
I have done some additional research and found a contact cement manufactured
by 3M that works for attaching the film to the insulating frame. Scotchgrip
#4693 is the stuff to use. You put a little on one or both surfaces to be glued
and let dry for 10-20 minutes. Then you put the two surfaces together and
Voila!, instant bond. The bond is so good that the film will tear long before the
glue lets go. Other contact cements may work well also. The only disadvantage
is that once you've assembled the driver using contact cement, you have to live
with it. If the driver doesn't work, you'll have to build another because you
won't be able to tear the old one apart.
7) High voltage DC bias supply (1000-5000VDC, almost no current)
This can be made as a voltage multiplier that works off the power lines. You'll
need high voltage diodes and capacitors, a few resistors, a circuit board and a
line cord. You can get away with one supply, but one for each speaker is easier
to deal with- you won't have to run high voltage wires all over your listening
room. See the Bias Supply section near the end of this document.
Optional:
Plastic coating for the perforated metal. I've heard that latex house paint works
fine...
Figure 1. Exploded view of a basic electrostatic driver.
Making the drivers:
Step 1. Design your drivers.
Decide on the size and make the frames for the drivers. It is generally easier to
make small drivers than to make big ones, but with small drivers you will need a
lot of them so mounting them can be a pain. 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 ft2 1500 V 1/16"
full range >4 ft2 3-5000 V 1/8"-1/4"
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.
Figure 2. 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
labeled "A", "B" and "C". Be sure to leave room for hardware to mount the
driver to some sort of frame.
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. That's why
we were careful to vacuum up the graphite powder. You may want to clean the
insulators with alcohol and a very clean cloth before proceeding.
The electrical connection is made by physical contact between a metal strip and
the graphite 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. 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 haven't
tried it.
I use a stretcher table of the type shown in figure 3. 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 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 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!
Figure 3. View of the underside of the diaphragm stretcher table. The film is
laid on the top side of the table and the edges are folded to the underside and
secured with double-sticky tape attached to the inside of the table edge. Inflating
the tube stretches the film tight. A rectangular table works just as well as a
round one and is probably easier to make.
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 anechoic 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? Time to
put the resistive coating on the diaphragm. First put the insulators in another
room. Then place a little (very little!) graphite on the film and grab a clean
cotton ball and start rubbing the graphite into the film. Rub it in hard. Add
more graphite as needed. You really don't need to use much. You want the film
to be coated with the stuff so that it has very high resistivity. It's really not
critical. After you have rubbed the graphite in, grab some clean cotton balls and
rub some more. You can measure the resistance of the film by dropping a
couple pennies on it a few inches apart and checking the resistance between the
pennies with a DMM. You want a high but measurable resistance. Move the
pennies around and check a few places. If you get resistances on the order of
100K or more, you've done a good job. If you measure lower resistances, rub
with clean cotton balls some more. Get out your vacuum cleaner, put a brush
attachment on it, and vacuum the entire surface of film that has been coated and
the area where you were using the graphite. Now wash your hands very
thoroughly! Then wipe the insulators with alcohol and a very clean rag to make
sure they are absolutely clean before proceeding.
Why is the resistance important? Sooner or later, a bug will get into your
speakers, or you will crank the volume a bit too high and your speakers will arc.
If you use a metalized diaphragm (low resistance) there is a good chance that
the entire diaphragm will flame out and you'll have to rebuild the speaker (but
it'll impress your friends!). If you use a high resistance coating, the amount of
current available to the arc is very small, resulting in a low temperature arc that
will at worst put a pin hole in the diaphragm. High resistance coatings that I've
tried do not cause the normally self-extinguishing polyester diaphragm to
become inflammable. This is another reason for using a very large resistance
between the diaphragm and the bias supply.
If you feel that you really need extremely high resistance for your speakers, try
using dish detergent or antistatic solution to coat the diaphragm. I have built
drivers using all three coatings and find no audible differences between them
(but maybe your ears are better than mine).
Attaching the diaphragm is easy. You simply put glue (Scotchgrip #4693) on
one of the insulators (again- don't completely cover the metal) and place it, glue
side down, on the coated film. The bond forms instantly, so make sure you set
the frame down on the diaphragm exactly where you want it. Once the glue has
set (after about 10 microseconds), let the air out of the tube and cut the film
away from the table along the edge of the insulator. Now turn over the
insulator/film assembly and set it back down on the table, diaphragm up. Coat
one side of the other insulator with glue, wait about 10-20 minutes, then set it
glue side down on the insulator/film assembly. Be sure to align the two parts
carefully before pressing them together- you don't get a second chance. You
might consider building some sort of fixture to ensure accurate alignment.
Now you can epoxy the perforated metal sheets to the insulator assembly. The
perforated sheets are made by running a roller with metal pins over the sheet
metal. That leaves the edges of the holes on one side rounded and the edges on
the other side sharp. Put the rounded edge side toward the diaphragm. Epoxy
the stators one at a time and be sure the epoxy has time to set before you pick up
the assembled driver.
I have done some experiments aimed at rounding the sharp edges of the holes.
One of the things I recalled from high school chemistry experiments is that
corrosion of metals occurs fastest at points of stress and sharp edges. I tried
using ferric chloride PC board etching solution from Radio Shack. Since
aluminum is more 'reactive' than copper I had to dilute the solution by cutting it
with water at about 1 part FeCl to 4 parts water. This kept the speed of the
reaction slow enough to allow me to observe progress of the reaction and
remove the aluminum when the edges were rounded. If you try this, be sure you
dilute the FeCl and then put a small scrap of aluminum into the solution to test it
before you put in the pieces you will use for your speakers. If you don't dilute
the solution you'll end up with a bad smelling, boiling mess!
Step 4. 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 figure 4. 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.
Figure 4. Diagram showing electrical connections to the ESL. Warning: the
voltage output from the transformer is high enough to hurt you! Be careful!
You may hear a whining sound. This is 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 stinks)! Don't do it!
Turn on the amp and play a CD. 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.
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 midrange 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 midrange is
weak.
What I am about to say will be regarded as heresy but before you believe what
you hear from people who claim to know everything about everything (I'm not
refering to Roger sanders here), remember all the suffering that has occurred
throughout human history because people blindly followed what they were
told by such self-proclaimed experts. Here goes: 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.
Section II ESL Bias Supplies
The ESL needs a high voltage DC bias supply. Generally speaking, the higher
the bias voltage you use, the higher the sensitivity of your speaker. However,
there are considerations beyond speaker sensitivity. If your speakers do not
have plastic coated stators, then 1500-2000 Volts is about the highest voltage
you will want to use, regardless of insulator frame thickness. Higher bias than
that leads to corona discharge and its attendant whining sound. If your speakers
use plastic coated stators, you can probably use higher voltages, but that will
depend upon the insulator frame thickness also.
If you use 1/16" thick insulators, then try 1500-2000 Volts first. If your
insulators are 1/4" thick, and the stators are plastic coated, you may be able to
use 5000V bias. For headphones, where the drivers are practically glued to your
ears, 500- 750V is probably adequate bias.
Bias supplies are generally made using a transformer to step up your local
power line voltage to 500- 1500V, then converting to DC using a voltage
multiplier/rectifier circuit. Voltage multiplier circuits are used in almost
everything that uses high DC voltages. You can find a good description of the
operation of such circuits in any edition of the ARRL Radio Amateur's
Handbook.
Figures 5, 6 and 7 show schematics of voltage doubler, tripler and quadrupler
circuits. The names doubler, tripler and quadrupler come from the fact that the
output voltage will be approximately 2, 3 or 4 times the PEAK level of the AC
input voltage. Transformers are usually rated in terms of rms voltages.
Vpeak = 1.414 x Vrms
