[SoundStage!]Max dB with Doug Blackburn
Back Issue Article
June 1998

Technobabble: The Importance and Meaning of Distortions

This month I’ll talk about a few of the technical aspects of the performance of high-end equipment that are misused and/or misunderstood or ignored. If you think you’ve been around high-end audio long enough that this will be too basic to bother with reading, allow me to encourage you to stick around anyway because I’m planning to drop a few bombs that will stir things up a bit. I’ve made up some of the technical terms too, just to add a bit of food for thought. There are things going on with (and within) your components that nobody is telling you about.

Let’s start with some discussion of distortion. People talk about it all the time, but do they really know what they are talking about? Simply put in electrical terms, distortion is deviation from perfection. In the case of high-end audio equipment, perfection is the incoming audio signal.

But let’s take a short detour from this paragraph for a second. Think about that last statement. "The input signal is perfection." If the input signal is from the headphone jack of a $50 portable radio/cassette player, is it perfection? Yes. The job of a high-end preamp or amp is to create an identical but larger version of the input signal. High-end components which output something different than the input signal are, by definition, inaccurate. Do you want inaccurate high-end components? Apparently some people do. Or else they do not yet know how to deal with the "accurate vs. musical" question. There is also the question of the goal of the audio system. Is the goal to reproduce the performance so that it sounds like it is happening live—as though you were there when the recording was made? Actually this has been espoused by some rather well-known, self-appointed high-end know-it-alls. If you tried to do this, you’d have so many components and wires fighting each other that you’d get nothing resembling pleasing music from your system. Besides, there are bottlenecks in the recording and manufacturing that make listening to a symphony at home and trying to make it sound like "live from Carnegie Hall" that the goal is ludicrous. The real goal of the high-end system is/should be to reproduce whatever is on the recording as well as possible. If the recording does not sound as "live" as if you were seated in the hall when the recording was made, so be it. If you tried to force the recording through components to make it sound "live" or "you are there," you would alter the original signal to a significant degree. This opens up so many pits to fall into that I’m quite sure that over the years people have started out with fine intentions and ended up trying to follow an impossible dream. You can’t do better than perfectly reproducing what is on the recording. Once you are "there," all you have to do is look for recordings made with the intention of making a representation of a real performance as if the listener were there when the recording was made. Now you put the responsibility for a certain amount of "performance" back onto the performer/engineer/producer—where it belongs. You can’t build a good system on the false hope of making recordings that do not sound live to come out of your system sounding live. Now, you can continue with the paragraph that was interrupted. In fact, here is the beginning of the paragraph again, with its middle and end:

Let’s start with some discussion of distortion. People talk about it all the time. But do they really know what they are talking about? In electrical terms, distortion is deviation from perfection. In the case of high-end audio equipment, perfection is the incoming audio signal. The component processing the audio signal is usually going to use the incoming signal as a template to create a new larger version of the signal. If the new, larger version is identical in every way to the original (except for size, of course), then distortion is "zero." The more the output is different from the input, the more distorted the signal is. There are many types of distortion. Let’s look at some common ones and some that are less obvious or completely unrecognized by high-end manufacturers and consumers.

Distortion or harmonic distortion measures how much "additional content" the output signal picks up inside the component. Let’s say we are testing the component with a musical signal close to 80Hz (Hz = Hertz = cycles per second = vibrations per second in air; humans hear sounds from 20Hz, the lowest bass, to 20KHz, the highest sounds; most adults do not hear frequencies higher than 14KHz-16KHz). Second-harmonic distortion for the 80Hz input signal occurs at 160Hz (2x80), third-harmonic distortion occurs at 240Hz (3x80), fourth-harmonic distortion occurs at 320Hz. It is generally accepted that third, fifth, seventh, etc., the "odd" harmonics, are less pleasant-sounding than "even" harmonics (second, fourth, etc.). However, even though even harmonics may be less offensive, they are still distortion. Some people have created products that intentionally add even-order harmonic-distortion products to make the sound "prettier." If you have one of these and enjoy using it, that’s fine as long as you understand what it is doing to the audio signal. Those products cannot make the music more accurately represent the recording. Tube amplifiers as a group have 10 to 100 to 1000 times more harmonic distortion than properly designed solid-state amplifiers. People used to blame the "bad sound" of solid-state amplifiers on having significant amounts of odd-order harmonics. Today’s well-designed solid-state amps probably do have a lot more odd-order distortion than even-order distortion. However, this is because the even-order distortion products are microscopic or possibly non-existent. The odd-order distortion in modern well-designed solid-state amps is far lower in magnitude than odd-order distortion products in tube amplifiers. (Note to rabid tube lovers: This is not a condemnation of tube components. It is merely a dispassionate statement of fact. We all know that a lot of people prefer the sound of tube preamps and amps—in spite of the higher levels of harmonic distortion. Again, that’s OK as long as we’re all on the same page about what’s going on.)

Intermodulation distortion – This type of distortion occurs when frequencies (sounds) occurring at exactly the same time during the music cause the component in question to produce some less-than-perfect representation of those two (or more) frequencies. You don’t hear too much talk about intermodulation distortion today. In the ‘70s it was a very hot topic, and the Japanese had a ten-year "race" to produce components with vanishingly low intermodulation distortion. Not that it helped their equipment sound any more musical! Today, most decent-performing high-end components don’t have much to worry about in the way of intermodulation distortion.

Phase distortion – The input signal is a constantly changing voltage—a different voltage at each instant in time. To be perfect, the component in question must not shift the output signal in time compared to the input signal. Each cycle (a complete positive and negative sequence) contains 360 degrees of phase. Since a cycle requires a specific amount of time (longer for bass frequencies, shorter for high frequencies), each degree of phase represents 1/360th of the time of one cycle. A 30-degree phase shift means the output signal is shifted in time by 8.33% of a cycle. A 30-degree phase shift is not a desirable thing at any frequency. Designing a component for minimum phase distortion is quite an undertaking since phase distortion tends to be frequency dependent. This means that in most components, there will be varying amounts of phase distortion depending on what frequency is being tested. A component might have zero phase distortion at several frequencies and 100 degrees or more at many other frequencies. Some speaker manufacturers pride themselves on their ability to create full-range loudspeakers with insignificant amounts of phase distortion. Many audiophiles purchase these products thinking they have done a good thing, then connect these nearly phase-perfect speakers to amplifiers with 30, 60, even 100 degrees of phase shift at some frequencies. It’s unlikely that these people will ever hear the benefit of that phase-perfect performance. It’s easy to end up with an amplifier that shifts phase quite a bit since phase-shift performance is not typically mentioned in spec sheets. Even if it were, most people wouldn’t know how significant it is. Likewise, if you are lucky enough to have acquired a phase-correct amplifier and you then connect it to loudspeakers which are not phase correct, you’ll never hear the full benefit of system-wide proper phase performance.

Making phase-correct loudspeakers is not for the faint of heart. This is fairly obvious when you start looking for speaker manufacturers who design for phase-correct performance. There aren’t many. This doesn’t mean phase performance isn’t important. It means that it is so difficult that most companies don’t bother trying. It also means that phase-correct speakers will tend to sound quite different than other non-phrase-correct speakers. And since most speakers do not sound like phase-correct speakers, people often assume the phase-correct speaker has something "wrong" with it, that it is "subtractive" or "lacks air" or "isn’t dynamic enough" or "isn’t focused well enough" or "is dark-sounding." A well-designed phase-correct speaker is going to be accurate. It’s the other speakers that have something wrong with their sounds because they are not phase correct.

Power response distortion – This is not an often-discussed type of distortion. In fact, I just made up the name so I could call it something. This kind of distortion may be a factor in some of the "unmeasurable" differences in components that audiophiles claim to hear but nobody claims to be able to measure. Most measurements of audio equipment are made at specific frequencies using one or two (simultaneous) frequencies. Sometimes a sweep (starts at 20Hz and progresses smoothly up to 20KHz without stopping) is used, and the response of the component is plotted for an infinite number of points along the sweep, so you get a smooth curve representing the response of the component. Ideally, there would be no variation in response from 20Hz to 20KHz. Some amplifiers can come very, very close to this ideal performance. However, something unusual happens in some components when you apply many frequencies at the same time, like with a symphony or "pink noise," which is a representation of all frequencies being reproduced simultaneously. These components do not remain "flat" when hit with a wide range of frequencies in a complex signal at the same time. Something happens to their response when they are stressed by a complex musical or test signal that is not obvious when using single frequencies or sweeps as test signals. This can happen to speakers or to active electronic components.

Group-delay distortion – This kind of distortion happens when different frequencies propagate through a component (or cable!) at different speeds. Different frequencies do have a tendency to want to travel on the same conductor at different speeds. We are taught in school that electricity travels at the speed of light. This is theoretically true, but it possibly never happens in real-world equipment as we know it today. In fact, most claims for propagation speed I’ve heard from manufacturers are between 60% and 85% of the speed of light. Light travels so fast that the difference between 60% and 85% is a very significant number. One cable manufacturer claims its newest audio interconnect propagates all frequencies at 95% of the speed of light. If true, this is probably as fast as you will ever hope to get in the real world without room-temperature superconductors. The problem with audio frequencies is the tendency for higher frequencies to travel faster than the lower frequencies. To make a product sound "coherent" you want all the frequencies traveling through the component to do so at the same speed. If there are differences, you end up with time delays at various frequencies that will impact the "feel" of the component more than other parameters audiophiles are used to talking about (like transparency, dynamics, etc.). Group delay also tends to introduce a sense of "smear" in the reproduced sound. Imagine writing your signature then putting your thumb on it and smearing the ink to the left or right a little. What you see as a result of this is rather like what you hear when this "smear" factor from group delay is large enough to be audible. Most people will say, "But I hear nothing like that in my system." Nor did I until I heard some components and cables without group-delay smear. It is a type of distortion that is not obviously audible until you experience its removal.

Transfer function – This is the term applied to the accuracy with which a device can amplify the input signal. A linear transfer function results in symmetrical amplification, which means the output waveform will be a perfect but larger version of the input signal. When the transfer function is non-linear, the output signal is not an exact duplicate of the input signal. Tubes have non-linear transfer functions, and there is nothing that can be done to change this characteristic. Designers of tube products claim to use "the linear portion of the curve" for the operating range of the tubes in their products. However, you’ll notice that the statement is a bit misleading. A curve has no "linear portion"—a curve is always curved. The family of curves for tubes is somewhat parabolic. There is an area that is sort of semi-linear, but it still has a curve to it. Transistors have and inherently linear transfer function. The "curve" for a transistor’s transfer function is actually a straight line until you reach the operating limit of the device. This linear transfer function does not automatically make solid-state amps sound better than tube amps; there are a huge number of other variables.

Roll all of these kinds of distortions together and you have a pretty complex set of limitations to have to work within. Frankly, many manufacturers either do not recognize that these are all important things to worry about or realize that they would never finish a product if they tried to address all of them effectively. But there are manufacturers out there who through sheer force of will, huge amounts of manpower hours, or long years of engineering knowledge and experience do manage to address all of these "nasties" to a significant degree.

Remember: If you go to the trouble of selecting one component that performs very well in all of the categories above and you drop it into a system where the existing components are all over the map as far as the other distortions go, you won’t understand the full value of your new carefully selected component. But as you replace components in your system with components that do better and better in terms of the distortions discussed, you’ll begin to notice something special happening. The music will start coming together—it will start to sound cohesive, organic, whole. And the closer and closer you get to the ideal of no distortion of any kind, the closer and closer you get to the original performance captured on the tape, record, CD or DAD—with the usual caveat: provided that the manufacturers of these components also know how to deal with all the other thousands of details that must be attended to in making a fine-sounding component.

Shopping for components which are low in all of the distortions discussed this month is not easy. You may have to do a lot of behind-the-scenes fact-finding to locate components that are intended by their creators to be low-distortion devices. In loudspeakers, time- and phase-correct performance are possible only when the speaker has first-order (6dB-per-octave) crossover slopes AND when the drivers are physically offset from one another to line up the voice coils. No dynamic (cone) loudspeaker I know of which has the drivers mounted on the same flat vertical front panel can be correctly time-aligned. A tilted-back speaker front might offer enough tilt to time-align the drivers, but not always. In amplifiers, it is very difficult to predict phase performance by looking at the amplifier or the spec sheet since these performance parameters are rarely mentioned. If you do find an amplifier or preamplifier claiming very low phase distortion, and graphs or other data are included to "prove" it, chances are fairly good that the designer really did aspire to low levels of phase shift.

...Doug Blackburn
db@soundstage.com

[SoundStage!]All Contents
Copyright © 1998 SoundStage!
All Rights Reserved