LOUDSPEAKER POWER REQUIREMENTS: In the September 1973 issue of STEREO REVIEW, Roy Allison presented a thorough analysis of the power requirements for realistic music reproduction in the home. Mr. Allison's article dealt primarily with the question of how much power is needed with typical low-efficiency speakers to give a satisfactory illusion of "live" listening levels without audible distortion from amplifier clipping. His conclusion was, in brief, that 40 to 60 watts per channel might be enough, provided the listener was satisfied with a maximum instantaneous sound-pressure level of about 100 dB. Under these conditions, amplifier clipping can be expected to occur regularly, but most listeners are able to accept the resulting distortion, which need not be grossly audible.
My own conclusions in the matter, reached during (and as a result of) test measurements and listening comparisons between a number of very powerful amplifiers, are consistent with the Alli son findings, although I am of the opinion that even the most powerful amplifier available may be inadequate for completely realistic sound levels under certain home listening conditions.
In looking at the problem once again, let us put aside most of the variables in the home music-reproduction equation and consider only the matter of the ratio of peak to average sound pressures in the acoustic output of musical instruments. (By "peak" I mean the highest levels attained by the audio signal, and by "average" the mathematical average of all the signals occurring over the same given period of time.) Studies published over forty years ago-and presumably still valid--indicate that in the output of most instruments the peak sound pressures (in limited frequency ranges) are typically 10 to 30 dB greater than the total average sound-pressure level (SPL).
Peak energy levels most often occur in the mid-range-from 200 to 500 Hz but there are many exceptions. Extreme examples include the pipe organ, whose sound peaks of 20 to 30 dB above aver age levels may occur anywhere in the frequency range from 20 to 2,500 Hz, and cymbals, whose peak energy is concentrated at the highest frequencies, with a maximum of +20 dB between 5,000 and 10,000 Hz. (Higher frequencies may have been involved, but 10,000 Hz was the upper measurement limit in the particular study in reference.) It should be noted that we are not dealing here with the total dynamic range of a given program, which extends from the highest peak level down to the no-signal noise level. That subject was treated in some detail at the recent Mid west Acoustics Conference (see STEREO REVIEW, February 1975), and the Conference panelists indicated some diversity of opinion as to the dynamic range desired or required for music reproduction in the home. Our concern at the moment, however, is with the upper portion of the program's dynamic range, which extends from the "average" level to the maximum peak levels, since this directly involves the question of loud speaker survival.
To return to those peak energy figures mentioned above, they appear to be large enough that a home music system would find it virtually impossible to re produce them unless they had first been subject to limiting or compression. To take the pipe-organ figure as an example, a 30-dB increase in sound level represents a power ratio of 1,000 to 1, which means that today's largest amplifiers would have to operate at average levels of a fraction of a watt to prevent clipping on peaks. The situation is saved, how ever, by two fortunate circumstances: our ears are able to tolerate without distress--or even recognition--consider able peak clipping (as long as it is not too severe or too frequent), and virtually all program material played through home systems has indeed already been processed to limit (compress) its dynamic range.
I have not seen any thorough study of the peak-to-average level ratio of re corded music (as distinguished from "live" music), but my guess is that it is in the 10- to 15-dB range for most classical music and that it rarely, if ever, exceeds 20 dB. Even if the latter figure is valid, it should be possible to listen at average levels of 1 watt or so with a 100-watt amplifier with virtually no clipping, and at substantially higher levels with acceptable amounts of clipping. On many occasions I have listened to music with an average power of 50 to 60 watts per channel using an amplifier that clips at about 250 watts (permitting a peak-to-average ratio of about 6.5 dB) and have not found the clipping distortion objectionable. Perhaps an SPL of over 110 dB is sufficient to mask the distortion that occurs on the loudest peaks, or perhaps is simply numbs the auditory system! The basic study of music-instrument peak-to-average output, of course, came long before the advent of rock and electronic music. Much hard-rock music has a very high average level, with the peaks rarely going more than 5 or 6 dB above it. In addition, distortion is frequently very difficult to hear in rock, whose sound may already be, in effect, "pre-distorted" by the techniques used in its production. Electronically synthesized musical sounds are not usually played at the extreme levels employed in much rock music, but they differ fundamentally from the sounds produced by conventional musical instruments. It is possible to synthesize almost any type of wave form, including essentially pure single-frequency square or sine waves. This presents the possibility of driving a speaker with a steady-tone signal, possibly at full power. Most speaker manufacturers caution against driving their products with such signals (especially at high frequencies) except at power levels under 1 watt or so.
All of this leads us to another important factor in the definition of "necessary and sufficient" power for home listening. No matter how much power an amplifier can deliver, there is a finite limit to the electrical input a speaker can accept without harm. A loudspeaker driver can be damaged in two ways: mechanical breakage of the cone, suspension, or voice coil, and thermal burn-out of the voice coil (which can sometimes be accompanied by a spectacular display of smoke and flames). The former type of failure is most likely to occur in the woofer, which undergoes large physical excursions in the course of its normal function. Some acoustic-suspension woofers are designed for cone movements of an inch or more, but any at tempt to drive them beyond that point usually produces enough audible distortion that it will effectively discourage the practice. However, an "accident" such as flicking the phono cartridge stylus with amplifier gain at maximum can result in a few hundred watts of power literally tearing the voice coil from its form or the form itself from the woofer cone.
The large size of a woofer voice coil, and the amplitude of its movement, provides enough cooling capacity to make thermal burn-out a rarity. Even so, a woofer designed to handle the high average power levels of rock music must have a voice coil able to dissipate that power for extended periods, and this calls for special high-temperature materials and bonding agents.
At the other end of the frequency range is the tweeter-by its very nature, a small and rather delicate mechanism.
The tweeter cone does not move visibly, and its voice coil is wound with fine wire in the interest of achieving low mass.
Physical damage to the cone or voice coil is unlikely, but thermal burn-out is an ever-present threat. To make matters worse, the tweeter does not provide early warning of impending disaster by sounding distorted: the entire catastrophe takes no more than a fraction of a second, and the tweeter is no more! The mid-range driver, whether it is a cone or a dome radiator, usually resembles the tweeter in its lack of appreciable physical movement and general fragility, although its voice coil is larger and more rugged. Unfortunately, many musical instruments deliver their peak energies in the mid-frequency range, and as a result the mid-range driver is usually more vulnerable to damage than any other part of the speaker system. As with the tweeter, its nemesis is thermal overload, and there is no advance warning that it is being overloaded.
Any speaker can handle large momentary power inputs if they are brief enough that the average power does not heat up the voice coil excessively-or damage the cone mechanically. As the duration of the input signal increases, however, the power-handling capability decreases. Most speaker maximum-power ratings are merely a rough guide to the amplifier rating that can drive them to a loud output without undue risk of damage. Some idea of the power-vs.-time relationship can be gained from the specifications for the AR-LST 1, one of the few speakers we have seen rated in this way. It can handle 23 watts indefinitely, 64 watts for 30 seconds, and 1.000 watts for 2 seconds. No distinction is made in regard to the frequency distribution of the input signal, possibly be cause the system's multiple mid- and high-frequency drivers provide a more or less uniform capability at all frequencies. We suspect that a somewhat similar relationship between power and time exists with most other speakers, al though the numbers will of course be different (and usually much lower).
Many speakers are protected by fuses (and, lately, by circuit breakers) either factory-installed or to be inserted in the speaker line by the user. Although fuse protection is always advisable, the degree of protection varies widely (I have blown out speakers without damaging their fuses!). If the fuse has too low a rating, or if it responds too rapidly to an overload, blown fuses become such a regular annoyance that one is tempted to forego their protection. And, too, a larger fuse or a "slow-blow" type may fail to protect the more fragile drivers. It is best to abide by the speaker manufacturer's recommendations in this respect-and better yet to avoid a gross mis-match between what the amplifier can deliver and what your speakers can take.
If you are aware of the risk, and are willing to assume it, most speakers can be used safely with amplifiers consider ably more powerful than recommended.
If you choose this route, do not try to play the speaker louder than would be possible with an amplifier of the correct rating. Let the added power give your system the "open" quality that comes with an absence of peak clipping in stead of risking speaker damage through an over-enthusiastic application of raw wattage.
==============
Also see:
EQUIPMENT TEST REPORTS: Hirsch-Houck Laboratory test results on the: Pioneer SA -9900 integrated stereo amplifier, Pickering XUV/4500Z stereo/CD-4 phono cartridge, Wollensak 4766 stereo cassette deck, and Tannoy Micro TM55DD record player, JULIAN D. HIRSCH
Dual cassette deck (ad)