Chapter 1 - Cone Speakers

I would like to tell you a few things about cone speakers that are not widely known. The cone speaker that we are talking about is a mass loaded device. It was developed by Rice and Kellogg in 1928 while working for General Electric. The purpose of the research was to develop a speaker that did not have the peaks and dips that speakers of this age did. What made this possible is that amplifiers were now available in the 1-watt instead of the milliwatt range. This advance brought us to the modern speaker. I would like to make the distinction that in a speaker we have multiple drivers. In general, we have woofers, mid-ranges, and tweeters. Each of this is a mass loaded device. They all work on the same principles even though they are different sizes, different shapes, and they have different frequency responses.

What determines the range of a cone driver is very simple. At the bottom end, the range is limited by the resonance of the driver and that resonance is determined by the mass of the cone and the springiness of the suspension. When you put that driver in a box the spring of the air adds to the spring of the suspension and that raises the resonant point. With woofers there is free air resonance and there is resonance in the box. It is the resonance in the box that we care about because that is what we are listening to.

A mid-range and tweeter also have a resonance at the bottom end, and at the top end all drivers are limited by cone break-up. For example, a 12” woofer may have a cone break-up that starts at about 1k Hz. A smaller driver will have a cone break-up that starts at a higher frequency. Break-up is determined by the material used in the cone and is caused by parts of the cone moving in while other parts of the cone are moving out. In effect, the cone itself is flexing. There are pictures of this where you can see ripples across the cone. If you can make the cone stiffer, then you can minimize this breakup. Of course, even if you minimize the break-up and you make a woofer that goes up to very high frequencies you then have the problem of dispersion. As the frequency gets higher and higher a large driver will have narrower and narrower dispersion. That is the other reason that mid-range drivers are smaller than woofers and tweeters are smaller than mid-ranges.

One of the advantages of a 3-way speaker is you can choose drivers and use them more within their ideal range. Essentially, we would like to get some distance above the resonance and some distance below the break-up because within this range the cone speaker has a very flat response. I would also like to point out that since the driver is a mass loaded device it also has a very constant impedance. When you see impedance curves that show a wide variance of impedance in speakers, or peaks and dips, that is often caused by the crossover. We will discuss the crossover and its role in speaker design in more detail later in this chapter.

There will always be a peak and impedance at the resonant point (the lower resonant point of any driver). Again, this is where the mass of the cone is resonating with the springiness of the air spring and the suspension. This rise in impedance is generally damped by the damping materials and so forth in the design. The rise in impedance presents no problem to an amplifier, but dips in impedance do present a problem. Dips in impedance are caused, generally, by the crossover. If one uses a 6 dB per octave or first order crossover there will theoretically be no dips in that when one speaker, the woofer for instance, is crossing out, meaning that its output is going down as you go up in frequency, the mid-range or tweeter is now coming in and it is now presenting its load. What happens at the crossover point is that the lower speaker is giving up its output and its impedance and the upper speaker, the next one in the chain, is coming in and taking over the program material and controlling the impedance.

A wise speaker maker would choose drivers of all the same impedance. They would choose either 8-ohm or 4-ohm drivers and consistently use those drivers throughout the speaker design. Quite often this does not happen. Partly this is because it is easier to make a tweeter or a mid-range more sensitive than a woofer. To make a woofer sensitive it takes a very large magnet and often a large cone and then often a large box. As a result, you might find that a designer will use an 8-ohm mid-range and tweeter and a 4-ohm woofer.

Crossovers are a big bugaboo in speaker design. The problem with crossovers, especially in a 2-way speaker, is that choosing the crossover point and choosing the slope can become a very delicate issue. In a 2-way speaker we take the woofer up as high as we can, and of course, we want to bring the tweeter in as low as we can. However, this presents problems in the selection of drivers. If you have a rather large woofer, it is very difficult to make it go above 1,000 cycles. It is also very difficult to find a tweeter that wants to go below 1,000 cycles. As a result, we have this space of a few thousand cycles that neither driver wants to cover. You have probably noticed that modern speakers are using smaller drivers now and even multiple woofers to have enough sensitivity at the low frequencies and yet be able to go up to high frequencies. Things like polypropylene cones tend to help this because they have less break-up characteristic.

Another way to solve the problem is to make a 3-way speaker. If this is done properly it can be a much better sounding speaker. One of the classic speakers is the Acoustic Research AR-3a which has a very broad range in the mid-range. The range is so broad you might sense you are hearing almost all the music out of the one driver. You could view this as the large woofer just assisting the low end of that driver and the tweeter assisting the top end of that driver. It is interesting to note that the original AR speakers mid-range went from 1,000 Hz to 7,500 Hz, and then later, in the 3a, Acoustic Research decided to drop that range to 575 Hz to 5,000 Hz. I believe they dropped the range because they wanted to get more of the lower mid-range in the same driver.

The crossover point of a speaker is a delicate matter in that if you crossover around 1,000 Hz you are dealing with this presence range from about 1,000 Hz to 2,000 Hz where the ear is most sensitive. There are several problems with crossing over around that point. One of them is that you are going from a woofer that at 1,000 Hz has very narrow dispersion, especially if it is a large woofer. Then suddenly you cross over into a tweeter that has very wide dispersion at that frequency. This dispersion difference will cause problems depending on where you are sitting, how close you are to the speakers, if you are off axis, and so forth.

Acoustic Research made a very wise decision to move that crossover point down and get it out of the presence range so basically you are listening to one driver. I might point out I owned these speakers and the mid-range driver in the AR-3a is an amazingly good driver. It is probably where most of the money went. That is a wise choice because the mid-range driver is the thing that we listen to the most. It is okay if a bass driver is a little distorted as it does not have to be as accurate as a mid-range driver. The same goes for a tweeter. If a tweeter has distortion the distorted frequency is up at a frequency so high that you would not even hear it.

This is a good time to talk about distortion in speakers. One of the problems with woofers is on the low end when you must choose a lower cutoff. If you want to make a cone driver, such as a woofer, go very, very low, you must add mass to the cone. It is as simple as that. It is a mass loaded device and if you add mass to the cone, you can drop the resonant point. Now, say you add enough mass to drop the lower frequency limit of the speaker by a full octave. You will then, without doing anything else, reduce the sensitivity of the speaker 12 dB. That is a very large change and would require an amplifier that was more than ten times more powerful to drive the speaker. It would also cause a problem in that the whole speaker would then have low sensitivity, and the mid-range and tweeter would have to be padded down. Padding down simply means that you have a driver in the system that is more sensitive than the others so you must bring the sensitivity of that driver down to match the other drivers. This is typically done with resistors. It ends up making the entire speaker less efficient.

As you know, the speakers in the 1980s were a whole lot less efficient than speakers are today. Manufacturers of drivers have learned how to make more efficient drivers and solve some of the problems of cone break-up. Cone break-up is the biggest limitation of the upper end of the driver. It is one that is very material sensitive, and it is really where the art of speaker making is at its best. The low frequency end of it really comes down to simply mass and how big of a magnet you are willing to put on the speaker. As an aside, I have talked to several speaker driver manufacturers and the magnet alone is the most expensive part of a cone driver. It accounts for more than half of the total cost of the driver.

Going back to the topic of crossovers, the problem is this. As I said before, a 6 dB per octave crossover is a gentle crossover. It results in a flat response, a flat impedance, and perfect phase between the woofer and tweeter. You can also do 6 dB per octave crossovers in a 3-way speaker. The other nice thing about 6 dB per octave crossovers is that they are the simplest in number of parts. The woofer crossover is simply an inductor (or choke) that is a coil of wire. The tweeter crossover is simply a capacitor, so we only have two elements. If you add the mid-range, you must add two more elements, another inductor for the bottom end and another capacitor for the top end.

Now, the reason that people use steeper crossovers is because they are taking each driver pretty much to its farthest point. For instance, they will want to take the woofer up just about to the point that you get into cone break-up. I will mention at this point that in less expensive speakers, and sometimes in more expensive speakers, the woofer inductor is eliminated. Part of the reason is because it is expensive, and what the manufacturer does in that case is simply let the woofer roll itself off. Woofers will roll themselves off into their break-up, but they will roll off rather irregularly with a lot of peaks and dips, so this is not the way to make a high quality speaker. The reason for using a steeper crossover, again, is to get away from the cone break-up. You must also consider the tweeter as it does not like to see low frequencies, so the more rapidly you can get out of the woofer and into the tweeter, the happier the drivers will be.

One of the next crossover slopes that is used is 12 dB per octave. That simply doubles the number of crossover components, but it has an interesting problem. If you wire the two drivers in phase, there will be an infinitely deep suck out at the crossover point. Suck out in this case means there will be no output of the speaker at that point. It is not always a terribly noticeable thing, but it is certainly not something that would look good on any kind of measurement curve, and it would be noticeable when the music is in those frequencies. It is very narrow, and I would point out that a dip in response, or a suck out in response, is far less noticeable than a peak in response. A peak in response gives a ringing type of sound, whereas a dip does not ring.

To cure the suck out problem in 12 dB per octave crossovers virtually every manufacturer simply reverses the phase of the tweeter. This would certainly bother a person who is worried about time alignment because you will never be able to time align a tweeter that is reversed out of phase. In fact, you might ask yourself, how good would a tweeter sound if it is out of phase? You can make a big deal out of time alignment in the advertisement, but when you come down to it there are many 12 dB per octave crossover speakers that sound just fine. Now going to 18 dB per octave, again, is another set of crossover components. So now we have three crossover components for the woofer and three for the tweeter, and if we add a mid-range, we must add six more crossover components, so now you are getting into a pretty complex crossover. The advantage of 18 dB per octave, of course, is that you are getting in and out of the drivers much more rapidly, so break-up is less of a problem, and now you can wire the drivers in phase without the suck out problem.

Today, the Linkwitz-Riley crossover is extremely popular. This crossover is generally 24 dB per octave and phase correct. It is also extremely simple to create this crossover. What they realized is that if you take two 12 dB per octave crossovers and you cascade them, one after the other, you will get 24 dB per octave. Since the 12 dB crossover reverses phase, if you do it again, you will reverse the phase again and so now it is phase correct. Although, you must realize you have gone around the circle 360 degrees, so think about that for a moment. If this 360-degree phase circle occurs, then we might ask what does that do to the time alignment? Although some people think it is very important, I feel that time alignment is not that audible. I would not use time alignment as a high criterion for choosing a speaker. While a speaker might be time aligned it could also have all sorts of other problems.

Now we should talk about the impedance of a speaker and the phase angle that accompanies it. In an ideal world, a speaker could have a very flat impedance curve and that impedance could be resistive because the nature of the cone driver itself is a resistive load with 0 phase angle. The phase angle becomes a problem, especially with solid-state amplifiers, if the phase angle is more than 0, such as being a capacitive phase angle or an inductive phase angle. In this instance the amplifier sees that as a short. The reason simply being that, with a phase angle, it means that the current and voltage are no longer in phase so that when the voltage is at a maximum, the current may be at a minimum. The real problem becomes when the current is at a maximum and the voltage is at a minimum because that is the definition of a short. Solid state amplifiers are designed to shut down or limit current into a short mainly because the transistors would be damaged very rapidly by a short circuit.

If you present the same kind of load to a tube amplifier it overheats the tubes and shortens their life, although the amplifier will still drive the load. Now when I see a speakers impedance curve and see a large phase angle, I know that is generally caused by the crossover because the drivers inherently have a constant impedance curve. So, if you see big peaks and dips in the impedance curve, it is probably the crossover. Now peaks in the impedance curve are not the problem, it is the dips. If the phase angle is rather high, say 45 to 90 degrees, but the impedance is also high at the same frequency, then the amplifier does not have a lot of trouble driving that load because again, the current will be low. The problem occurs when the current is high, and the voltage is low which causes the amplifier to limit current. Therefore, keep in mind, if you purchase a speaker that has a large phase angle and low impedance at that phase angle you will need to find a very high current amplifier, preferably one without load line limiters. These amplifiers tend to be more expensive because they must use a greater number of transistors.

Another advantage of having a flat impedance curve is amplifiers with a low damping factor will be presented with a resistive load and will not alter the frequency response of the speaker. If an amplifier has poor damping that means it has a bit of output resistance and the output resistance interacts with the impedance of the speaker causing the response to be different than the manufacturer intended in the design. In John Atkinson's reviews of speakers and amplifiers he points this out using a standard variable impedance and seeing how the amplifier will influence that impedance, or better said, how that impedance will influence the output of the amplifier. We want the output of the amplifier to remain flat with varying impedance.

It is now time to talk about how you put a speaker in a cabinet. If you have ever experienced a cone speaker out in the open, you will notice it does not have any bass response at all. This is because the back wave comes around and cancels the front wave at low frequencies and mid-frequencies and only the highest frequencies will be short enough wavelengths to be blocked by the back of the speaker. The solution for this is to put the speaker in a box or in a very large baffle. A box is preferable because even in the large baffle it is very hard to make the baffle large enough to block the lowest frequencies when one keeps in mind that even at 50 Hz the wavelength is 20’ long. In essence, to even play it out to 50 Hz one would need a baffle of at least 10’ in size. In that case we are talking 10’ x 10’. An ideal baffle, by the way, has long been to just put the speaker in the wall and let the back wave play into another room, or a closet, or a chimney in some cases. That is the definition of an infinite baffle.

Since an infinite baffle is not a practical thing especially to sell in the store so we need to come up with another solution. There are two solutions to the back wave problem. One is called the sealed box and the other is a vented enclosure. The vented enclosure will extend the bass response a bit lower than a sealed enclosure. In fact, it was not until the late 1950s that Edgar Villchur developed the closed box system. It was quite a different thing, and it really shook the audio world that somebody would come up with this idea. It was so revolutionary it required a special woofer because woofers of the day did not have enough mass to work in a sealed box.

The advantage of a sealed box is that it has a very smooth roll off at the low resonance and rolls off at precisely 12 dB per octave. The disadvantage is that to get the frequency response to go very low (and get the resonance at a very low frequency) one must make a driver with a heavy cone which as we now know reduces the efficiency. In addition to heavier cones, we also need a larger box. One will note that modern sub woofers that go very low might have a small box but have an incredibly heavy cone, and this heavy cone is often driven by kilowatt amplifiers.

The vented box speaker is popular because it can be a little bit more efficient, and it can go a little bit deeper in bass response. The way it does this though happens to be something that I do not particularly like. To achieve this extra half an octave or so one basically creates a resonator that is known as a Helmholtz resonator, that is the same principal by which an ocarina works. The earliest vented speakers were done with an open port where they just had a square or rectangular opening usually below the woofer. Later, it was decided that one could make the port by using a round tube of a certain diameter and a certain length.

However, there are two problems with the bass response of a ported speaker. One is we have now created a resonator and we know in a way that resonance is not something we desire in high fidelity. This resonator will have a bit of a peak at this resonant frequency and create, to some extent, one-note bass. However, the other problem is that now the low end of the speaker starts rolling off at 24 dB per octave instead of the 12 dB per octave of a sealed box. If you are listening to tones that are very, very low, such as 16 Hz to 20 Hz these tones will be much farther down in response with a ported speaker because it drops off much more rapidly.