Chapter 2 - Electrostatic Speakers

The first thing I would like to say about electrostatic speakers is that they should not be feared as much as they seem to be. There are stories about speakers arcing, being over driven, and burning up, but this is really a rare occurrence. If you buy a proper electrostatic speaker and a proper amplifier to use with it, you will not have any of these problems. The most famous electrostatic speaker is the Quad ESL (also referred to as the Quad 57). While it is known to have a wonderful mid-range, one of its problems is that the bass response is rather poor. The bass peak at 90 Hz is very narrow (3 Hz wide) and very tall (~20 dB high). However, were it not for that peak, the speaker would have no bass at all.

So now is a good time to discuss why electrostatic and magnetic panel speakers have a problem with bass. One of the reasons is because you are dealing with a speaker that has an open back and very little baffling. These speakers, in general, are only about 2’ wide and they are often 6’ tall, which is generally a good thing, since height is needed if you want a line source. By definition, a line source should go from floor to ceiling. However, it is usually adequate for a line source to go up to 6’ tall or so if you want to stand up and listen. For sitting and listening, you can get away with a line source that is only about 4.5’ tall. Of course, when you stand up, you will hear the treble drop off because a true line source has very little vertical dispersion. It does not project anything above or below its line. The virtue of a line source is wide horizontal dispersion that projects very well to either side.

Another reason as previously mentioned is that electrostatic and magnetic panel speakers will have a bass peak and if you do not have that peak, there will not be any bass. The bass peak will be at a specific frequency and this frequency is different for every speaker. The problem with a bass peak is it results in what we call one note bass. Any note played near that frequency will be accentuated, while other notes will not be heard as clearly. For example, when a walking bass line is being played, you hear certain notes very loudly and other notes not very well at all. The best way to handle this problem is to add a subwoofer.

I know there are people who think there is no way to integrate a subwoofer with an electrostatic speaker, but indeed there is. One of the things to consider is taking the low frequency material away from the speaker. If you let it stay there, then you still have the bass peak, and the subwoofer will have to work against that peak. The solution is to use an electronic crossover so that you can split the signal before you go into the amplification. Now there are also people who object to the electronic crossover. I would like to make a point about that. An electronic crossover is merely a line amplifier with some very simple filters that use resistors and capacitors. So, it is not something that is going to degrade the sound. Although, I am not saying that the electronic crossover in an active subwoofer you might buy would be that good.

What you really want to do in this case is buy a separate electronic crossover that takes the signal from your preamplifier and then routes it to two power amplifiers. One amplifier for the mains speakers and one amplifier for the woofers. You will always benefit in the fact that the woofer amplifier can be a decent inexpensive Class AB solid state amplifier that puts out 100 to 200 watts. The amplifier does not have to excel in the mid-range or top end because it will not be playing that program material. It will certainly be better than any plate amplifier you would find in an active subwoofer. In fact, I have to say that I do not think much of active subwoofers, and it is a bad idea to put an amplifier in a woofer box. Frankly, as I have serviced subwoofers, I think the woofer shakes the amplifier apart and over the years causes solder joints to fail.

Not all panel speakers are line sources. In a panel line source, you must portion the diaphragm vertically into small strips. My electrostatic speaker is designed this way, as are the Quad ESL and Magnepan speakers. Roger Sanders does not segment his driver, so his speaker has very narrow dispersion. Some people like narrow dispersion, but narrow dispersion is like sitting in a room with two spotlights aimed at you. It does not illuminate the room very well at all. Harold Beveridge achieved 180-degree dispersion at all frequencies by using an acoustic lens. His speakers had a 12” wide and 6’ tall single diaphragm playing full range music. The lens is a time delay device that converts the flat wave into a half cylinder wave giving you 180-degree dispersion. Now to get wide dispersion in a panel speaker, the tweeter section must be very narrow, say 3/4”. If you do a three-way version, you have a mid-range section that is maybe a few inches wide. Then the bass panel can be as wide as you want as you are crossing all these segments over at different frequencies, and in doing so, you choose frequencies that are commensurate with the width of the driver. My speaker has five different segments and I get very smooth response and very constant dispersion over a wide range of frequencies.

At this point, you may want to know why a wide speaker, whether it is a round cone speaker or an electrostatic or magnetic panel speaker, narrows its dispersion as you go up in frequency. If this did not happen, then we can certainly make a speaker with a single wide diaphragm and still have wide dispersion. The reason for this is quite simple. At 10k Hz a wavelength is about 1”. Now if you have a driver that is many, many inches wide, then you have something that is several wavelengths in width. When the wavelength of sound becomes so short that there are many wavelengths across the front of the speaker, the dispersion becomes very narrow. To get very wide dispersion, you make the driver close to one wavelength or even half a wavelength in width. This is part of the reason why tweeters must be small and mid-ranges a little bigger. Woofers are large because we need to move a lot of air.

Lord Rayleigh back in the 1800’s wrote the equations and drew curves for what a speaker would do, which is quite amazing as he did not have a speaker to test out his theories. However, he wanted to know what happens when you put a 1” piston in an infinite baffle. Imagine a 1” piston vibrating in a wall. What happens as you sweep the frequency? What happens to dispersion? What happens to the output? The answer in a practical sense is that we could make a 1” piston speaker and it could play all the way down to the lowest frequencies, but the problem at those frequencies is the displacement or linear in and out motion of this piston would be enormous. This piston might have to move 6” or 12” to produce reasonable sound levels and very low frequencies. That is not very practical, but if we could figure out a practical solution that would be an excellent way to make a speaker.

Now since I mentioned the 1” piston moving at very large distances, the treble would also be riding on this 1” piston, and you get what is called Doppler distortion. The Doppler effect occurs when a train is coming toward you and the whistle is at a certain frequency. As the train is leaving, it will be at a lower frequency. Doppler distortion does the same thing. If the treble is riding on a moving woofer, you get the Doppler distortion of that woofer moving. That is another reason for making two and three-way speakers, because now the higher frequencies are on diaphragms that are not moving far at all. The same thing happens in a panel speaker.

One way to think about panel speakers is that they are tweeters, mid-ranges, and woofers that are stretched out into long strips. Of course, the beauty of the long strip is that it is a line source. I would like to add that a line source has another interesting effect. A problem we have with small speakers mounted on stands, or even large speakers, is the floor bounce effect, where you have sound coming directly toward you on one path, then you have sound going down to the floor and bouncing up at you on the longer path. The result is the two paths have different arrival times. Something to think about when listening to a line source is that you are listening to the portion of that line source that is directly across from your ear. For example, if you are sitting with your ear at 30” you are only listening to that part of the line source. The rest of the line source is merely supporting that part to complete the length of the sound wave so that you do not have the floor bounce. With any good line source, you can move your head up and down from floor to top of the panel and there is no change in sound.

Now a true ribbon speaker is very much like a true ribbon microphone. The way a ribbon is constructed is you must make two very long and very strong vertical magnets. Then you suspend a narrow ribbon of aluminum foil between them. Typical ribbons are only a 1/4” to 3/4” wide, because if you try to make the opening wider than that, you cannot get very much of a magnetic field. When you bring magnets very close together, they are very strong. As you take them apart from each other, the field gets much weaker. Even if you use very high-density neodymium magnets, it is still very difficult to get a strong field over a large distance.

You will see ribbon speakers such as Magnepan, where they are using the ribbon for a tweeter, which is appropriate, because you want something very narrow. The Magnepan tweeter is about a 1/2” wide ribbon. Then if they wanted to go further, they could make a wider ribbon for the mid-range. However, I believe after the tweeter, Magnepan goes to their standard construction, which is a wire glued to a Mylar diaphragm that operates in an entirely different way. There are some who think Magnepan speakers are electrostatic, but there is nothing electrostatic about a Magnepan. It is truly a magnetic speaker where the tweeter section is a ribbon, and the lower frequencies are a planar diaphragm.

The Heil air motion transformer was very famous years ago. ESS and some other manufacturers used them. The Heil transformer is very interesting in that the diaphragm is a folded Mylar diaphragm with aluminum deposited on it. As Oskar Heil used to say, it literally squeezes the air out of it. However, if you think of a piece of ribbon vibrating in and out, you are not really squeezing the air, you are pushing the air. With the Heil transformer the folds in the diaphragm come together and spread apart as you play it. It squeezes air out and sucks air in, so that a rather small unit can produce quite a bit of sound. That is one of the advantages of it. Some people were concerned about the fact that if you look at the diaphragm, it appears rather irregular. They start to wonder about linearity and if it has low distortion. I believe the modern ones are doing a very good job with this. Again, because it is rather small, you are not able to take the air motion transformer to very low frequencies.

Now a ribbon, even if it is a 1/2” wide and 6’ long, has very low impedance as you are dealing with the resistance of that piece of aluminum foil. There really is no motional impedance part of the equation. Magnepan has been able to make the ribbon thin enough to get the impedance up to 3 or 4 ohms. Shorter ribbon speakers (because you cannot make the aluminum foil but so thin and have it not tear) such as the Decca ribbon tweeters, which are only a few inches in length, would have to use a transformer because the impedance of the ribbon alone is only a few tenths of an ohm, and so using an impedance matching transformer gets that impedance to 4, 8, or 16 ohms. At those high frequencies a transformer can be rather small and even though some people have tried removing the transformer and driving the ribbon directly, you are probably just as well off using the transformer and driving it with a conventional amplifier.

Another type of speaker that is just a tweeter is the Ionovac, which was also known in the U.S.  as the DuKane. It is an ionic tweeter, and they had a very high frequency oscillator which used a vacuum tube. This oscillator would produce a cloud of ionized air in a very small chamber. DuKane had to use a tungsten electrode and because it ran red hot, they put quartz glass around it. By creating this ozone and then modulating the intensity of it, you could create a tweeter. This is basically called a plasma tweeter. There was also the Heil plasma tweeter, which used a tank of helium, which is a bit inconvenient. I am not quite sure why Oskar Heil decided to use helium, but it is basically the same concept. You are modulating an ozonated or plasma of air. You can vary the intensity of that plasma by modulating the oscillator and make a very fine tweeter. The Ionovac tweeter because of its very small generating size, was coupled to a horn, which then loads the device. It was a very popular and good sounding device.

Now I would like to go into a little bit more detail about the electrostatic speaker, because I feel the electrostatic speaker is one of the finest ways to make a very good sounding audio loudspeaker. I have often said to people if you can make an electrostatic speaker and get it to work at all, it will probably sound pretty good. Of course, this depends on how one handles the dispersion problems. Most people who make electrostatic speakers tend to make them with flat panels or try to curve the panel. By curving the panel, you can correct the dispersion problem, but you still have the problem of driving the diaphragm full range. I prefer driving the diaphragm in segments and many other electrostatic speaker designers have done this as well.

The basic operation of the electrostatic speaker is quite simple. You take a Mylar diaphragm that is very thin, say a half mil, and you stretch it tight in a frame. Then on either side of this, you put a perforated electrode. This electrode can be a piece of aluminum with holes in it, a wound wire electrode, or a fiber glass circuit board type of electrode with copper facing on it. These have all been used by different manufacturers. If we want to talk about the particulars, Acoustat and Roger West used wound wire electrodes. Beveridge used a cast electrode that was made from some very exotic materials. Martin Logan used a curved metal electrode.

One of the problems with curved electrodes is you must stretch a flat diaphragm into a curve. Since you cannot stretch something into a curve, what they must do is put several horizontal ribs across the curvature to try to hold the diaphragm away from the electrodes, but the diaphragm will never be equally spaced between the electrodes. That is a bit of a problem. Plus, the diaphragm must be stretched very tight in the vertical direction and not very tight in the horizontal direction. Now if you make a flat panel electrostatic speaker, and you segment the diaphragm, the next problem you have is how do you drive the speaker. An electrostatic speaker in general is going to have a charge on the diaphragm that is basically a static charge. It is supplied through a very high mega ohm resistor. The diaphragm is coated with a very light wash of some slightly conductive material such as a carbon wash. Quad used a water absorbent nylon, which means that their diaphragm resistance changes with humidity. Sometimes it takes a long time to charge up the diaphragm and other times it charges up very quickly. Others like Acoustat used a heavy carbon wash and it charges up very rapidly. The carbon wash is probably the most popular way to put a charge on the diaphragm.

We use a carbon wash because we do not want a fully conductive diaphragm such as aluminized Mylar. The reason for this is that as the diaphragm moves and comes closer to one or the other electrode, the charge will want to scoot away from the closest part of the electrode, and we do not want this to happen. We want the charge on every part of the electrode to stay in the same place, much as it would on the surface of a rubber balloon. Then you must put the audio onto the stators of the speaker. This is also going to be very high voltage audio. Beveridge, Acoustat, and I made very high voltage amplifiers that run the range of 3,000 to 5,000 volts of audio so that we can directly drive the stator. The virtue of this is many folds as it is very difficult to make transformers that step up the voltage of a conventional amplifier to the levels of electrostatic speakers. The step up ratio of such a transformer needs to be about 200:1. It is very hard to make a 200:1 step up transformer that also maintains 5,000 volts of insulation. Frankly, you must maintain more insulation, because somebody may connect a very large amplifier and stress the transformer to higher voltages.

The main advantage of electrostatic or any panel speaker is now that we have a very large, segmented diaphragm, as we are driving this diaphragm the segments are completely linear over the surface area. Another way of saying this is we are putting the force on the diaphragm in every place as an equal small unit of force, as opposed to a cone speaker, where you have a very large diaphragm, but you are driving it from a very small central point. In a typical 12” or 15” speaker, the voice coil, which is where the energy is coming from, might be 1” or 2” in diameter, and you are driving a cone, which may be 12” or 15” in diameter. Now when you drive a cone of that large diameter, you will get flexing of the cone at certain frequencies. This is what is called cone break up. In an electrostatic or panel speaker, you are free from cone break up. That is one of the greatest advantages of such a speaker.

Another advantage is you are in a much more linear region. In the plastic electrostatic, which we call the constant charge electrostatic, the force on the diaphragm is directly proportional to the voltage applied, but the beauty of this is that no matter where the diaphragm sits in the gap, the force is still linear. This is not true of all electrostatics. If you make a conductive diaphragm electrostatic such as Beveridge did, the sensitivity of that diaphragm does vary with position. Bev’s feeling was that in his speaker, the diaphragm motion was so small, and it was not a push pull configuration, and so there was some cancellation there. He felt the distortion was extremely low, and indeed, measured it was quite low. The Beveridge speaker is one of the very few that had a conductive diaphragm. The rest of the speakers you will be familiar with are all, what we call, constant charge. That was really pioneered by Peter Walker of Quad.

Now I would like to come back to the Quad ESL. Many people do not realize this is a three-way speaker. The tweeter is about 1.5” wide. The mid-range, which includes the tweeter, is even wider at about 6”. Then you have two woofers on either side of that and they are about 12” wide, making the whole speaker about 2.5’ wide. Now Walker wanted to make a floor mounted speaker that would fit in a small English room. So, to get vertical dispersion, instead of making the speaker tall, he curved the panels. What Walker did, very cleverly, is make the panels as a flat unit. Then because they were made with a plastic fader, he curved them into a rather gentle arch. The arch is only about 1.5” or 2” from middle to end. It is not a great amount of curvature and not enough to displace the diaphragm any great deal.

Now the Quad 63 is an entirely different speaker, and the basic design is still being used today, although many versions removed from the Quad 63. It is just that the larger Quads have additional woofer panels to get more bass. The essence of the Quad 63 is that it is a true point source. Now making a true point source is a very difficult thing to do. The way Peter Walker went about doing it is he made concentric rings that expand out to be about 14” in diameter. This is in the center of the speaker. By making a time delay network, he delays the audio to each of these concentric rings, so that the outer rings speak after the center ring. Another way of thinking about it that the very center of this speaker speaks first. As that is coming toward you, the outer rings speak. They are all speaking at the same frequency, the same material, but the time delay creates sort of a half basketball of sound coming toward you.

Now my only complaint about the speaker is the fact that this point source makes you very aware of the center and as such you would like this point source to be up at ear level. Many people have taken the point source Quads and put them on tall enough stands to get them up to seated ear level. This is part of what Quad is now trying to do with their taller series of speakers. I really feel that if you do not bring a point source, no matter how the point source is made, up to ear level, you are too much distracted by the fact that the singer or the player is a very short person sitting on the floor and not up at a realistic level. I think it is very important with a point source to have the point at a reasonable physical level so that it is a believable sound stage.