This was It is with some trepidation that I bring up the issue of distortion in loudspeakers, but I really feel that this issue has not received enough attention. Loudspeaker manufacturers seem to think that if they do not mention distortion it will go away. The loudspeaker is by far the poorest link in the reproduction chain for all forms of distortion, and this includes nonlinear distortion, frequency response distortion, phase distortion, and time distortion. Compared to the electronics of a system, loudspeakers are horrible. As the recent directory issue of Audio shows, there seem to be a thousand different loudspeakers, only a few of which have the quality and loudness capability to approach realistic sound levels with the low distortion levels exhibited by even very ordinary electronic components. Clearly, this situation exists because most people want small, inexpensive boxes. So let us for the moment forget about what the market demands and discuss what loudspeakers actually give and can give in terms of distortion. Though there are many important design factors in making a good loudspeaker, I want to present data and make a few comments about one aspect of design: the distortion generated by the woofer at low frequencies and relatively high output levels.
It is fairly difficult to find data in the literature and in test reports that give consistent and comparable information on loudspeaker distortion. Some reasons are that it is difficult to make such measurements, there are no standards for frequencies and loudness levels, and manufacturers are certainly not going to give distortion figures, even when they know them, because the numbers are so high. So, with motto in hand, "non illegitimacy carborundum," I have gathered as much information as I could find over the past five years and compiled just a few sample cases for this note. Most loudspeakers of a given size are about as bad as one another. The list contains a spectrum of commercial loudspeakers, but unfortunately, no samples of large profession-al systems, which are very likely much better than most of those listed. I have found data on about 200 loudspeakers, from which I have selected 19 for the accompanying table. Here is some of the bad news about loudspeakers. The distortion data is taken at about 40 to 50 Hz and at three loudness levels: 80 dB SPL, 90 dB SPL, and 100 dB SPL at one meter on axis. Only the second and third harmonic distortion levels are shown and are given in percent distortion with respect to the fundamental frequency. As is apparent, most loudspeakers will not reach a level of 100 dB SPL without such severe breakup and distortion as to make the measurement meaningless. Bear in mind that 90 dB SPL at one meter is not very loud at all. For a medium-sized room, this would correspond to just about that level with two loudspeakers going (stereo) at the listening position. This level is barely satisfactory for orchestral music on loud to average passages, and no allowance is made for headroom at all.
In some cases, I had to estimate the data and adjust it to insure consistency over the
whole array of data. Thus, these figures should not be taken as precise, but rather used to get an ap-proximate idea of how bad the distortion problem really is for loudspeakers at the lower end
of the audio spectrum. An overview of the data seems to show that some of the better
loudspeakers will deliver about 90 dB SPL at about 1 percent distortion at about 45 Hz. The distortion
rises rapidly as the level is increased or the frequency lowered even a small amount. Because loud-speakers like amplifiers, should have some level headroom, so that the peaks are not
totally smashed, it should be clear that the sample speakers listed do not for the most part
reproduce sound at realistic levels without considerable distortion. Surveying this data is very
discouraging. On the other hand, no data is presented, or available, for the really large systems that
are now available. I have included distortion data for my own large system as a reference point
to indicate what can be done if the low-frequency distortion problem is seriously attacked by
brute force methods. In a number of A-B comparisons I have made between my reference system
I suppose most people have listened to a long string of high-distortion loudspeakers for so long that they think music is supposed to sound like hash when the level is turned up a little. In fact, it should not, and it does not. Live music sounds just fine at high levels, and there is no reason that reproduced music should not also sound fine. In fact, it does when the distortion from the loudspeaker is kept well below one percent. When the distortion in the system is lowered, it is possible to raise the listening levels to realistic values and still be pleased with the quality of the reproduction. I see no reason why distortion in loudspeakers should be any less important than that in the rest of the chain of reproduction. I am fed up with individuals who claim they can hear the difference between a power amplifier with 0.05 and 0.01 percent distortion while ignoring distortion that is 10 to 100 times greater in the loudspeakers. For the most part, I believe these people are not saying anything about the quality of the equipment, but rather are simply giving their opinions about what kinds of distortion they like.
In looking over data on hundreds of loudspeakers, there seems to be no good correlation between low distortion and design factors, such as acoustic suspension, open box, ported box, and the like (with the exception that transmission lines are much worse). There is a definite correlation between piston size and low distortion, however. Generally, the larger the piston (total piston area), the lower the distortion. Even this correlation is not perfect, because some manufacturers have demonstrated that they can make large, but very poor loudspeakers.
I have wished for years that designers would pay more attention to lowering the distortion in loudspeakers than to making them small for the sake of smallness. It appears that there is some hope in the recent appearance of new and technically improved designs in some of the larger systems. Though even the best of these has considerable distortion, they will start to educate the ears of the listeners to the higher quality that is possible with such designs.
I also hope then one day some brave loudspeaker manufacturer will meet the challenge, and have the nerve, to publish meaningful data on his loudspeakers. This would include items such as a meaningful power rating, distortion across the frequency spectrum (at several levels), efficiency, frequency response, phase response, and other information of the kind now published for power amplifiers. Perhaps then the quality of the loudspeakers would improve and someday even come within a factor of 10 of the best electronics.
(Ed. Note: Recent spectrum-level data taken during actual symphony concerts shows that the maximum levels in the one-third-octave bands at or below 50 Hz, which occur during works with loud bass drum such as the Tchaikovsky Sixth Symphony or the Rite of Spring, are typically around 94 to 96 dB SPL at the podium and 87 to 89 dB in the audience. The data will be published in the Speaker in the next few months. As for data from manufacturers, AR and Allison Acoustics, to name two, have for years been publishing frequency response curves for both individual drivers and systems, as well as distortion data at low frequencies as a function of level.)
Note: The last loudspeaker is not commercially available. It is the author's personal reference system, using four JBL LE 15 A drivers in an 800-literbox for each side. It will deliver 110 dB SPL at 40 Hz with 3 percent THD.
Enjoy the Music.com highly encourages our readers to join the Boston Audio Society by clicking here).
This article is copyrighted © by the author or the Boston Audio Society.
All contents copyright© 1995 - 2013
Enjoy the Music.com®