E. Brad Meyer worked at the acoustics firm of Bolt Beranek and Newman for six years, where he specialised in sound and vibration measurements and data reduction. Through his company, Point One Audio, Inc., he has been a free-lance consultant, writer and recording engineer/producer since 1976. His writing has appeared in, among other places, High Fidelity, Stereo Review, dB Magazine, The Boston Phoenix, Stereophile and Design News. In the early 1980s he was editor of the Boston Audio Society Speaker, and is currently president of the Society. He is an Executive Committee member and former Secretary and President of the Boston Section of the Audio Engineering Society.
'Power - How Much Do You Really Need?'
Hi-fi experts, like psychiatrists, have the annoying habit of answering questions with other questions. If you ask a psychiatrist, "How much power do I need?" he'll say, "How much power do you want?" The audio maven is much more annoying because he comes back at you with a whole series of questions: What kind of music do you listen to? How loud do you like to play it? What kind of speakers do you have? How big is your listening room, and how is it furnished and decorated?
All that information is needed not just to fine-tune a reply, but to give any sort of meaningful answer at all. Variations in the desired loudness level, in the choice of speakers, or in the nature of the room and its furnishings can each change the system's power requirements to a huge degree. (The psychiatrist needs little specific information; in his field the giving of answers is generally regarded as Bad Form.)
A man I know who is both a psychiatrist and an audiophile reports that someone once said to him, "What do I need a bigger amplifier for? Mine already plays loud enough to hurt my ears!" Of course, he pointed out that a big enough amplifier will play loudly without hurting your ears; but how do we know how big is big enough?
The table below will give you an idea of the nature of the problem. It shows the amplifier power required to achieve playback levels equivalent to various kinds of live music. The conversions from electrical to acoustic energy were made with the aid of formulas and tables from chapter 10 of Beranek's "Acoustics"; the sound levels come from my own measurements. The power figures are expressed in watts per channel, and are based on the assumption that there are two channels playing. Here's the table, along with a few important points to remember when looking at it:
Click here or on the image above to view the full table in this window, or here to launch the table in a new browser window which allows the table to be easily referred to whilst reading the article.
The most important thing in the table is how fast the numbers change in the vertical direction. Human hearing is logarithmic in nature, which is the main reason the question of power requirements is such a tricky one. An increase in level of 10 dB, which most listeners rate subjectively as "twice as loud", requires a tenfold increase in amplifier power. Conversely, if you are willing to play music at a level that is subjectively only half what you would encounter in a concert hall, your requirements decrease by a factor of ten.
What does this mean when you're choosing an amp? First of all, amplifiers differ less than you might think. Going from a 50-watt/channel model to a 100-watt unit will increase the available output by only 3 dB, which is noticeable, but just barely -- assuming the amplifiers perform similarly within their ratings. (You can't always assume that, of course; see the section below on output current.) Second, for someone who never listens above moderate levels, 20 watts per channel will be enough for even the most inefficient speakers.
Third, the average power required for most systems, even playing loud classical music at concert levels, is very low. It is the momentary peaks that require a big amplifier. The difference between the average and peak levels on classical material ranges from 8 to 12 dB, and although a well-designed amplifier can reach its maximum output level and "clip" occasionally with only a mild effect on the sound quality, an amplifier with generous headroom -- the ability to put out greater than its rated power for short periods -- has a real advantage over another one with the same continuous power but less headroom.
Rock music is different. The 116 dB given as the maximum peak level represents what you'll encounter at an actual concert. But in most rock the peak and average levels are very close, so that the corresponding average level would be around 110 dB. Most people can't play their systems at those levels because neither their neighbors nor their loudspeakers will stand it. If you must have such levels despite the inevitable hearing damage they will cause, you should achieve them not by buying a bigger amplifier but with special loudspeakers, which is our next subtopic.
The differences between the columns in the table reveal that your choice of loudspeaker can change the power requirements by a factor of at least twenty: a speaker that puts out 99 dB at 1 meter with one watt of input will play as loud with a 3-watt amplifier as an 81-dB/w/m speaker with a 200-watt amp. If you must hear classical music at realistic levels, it's pretty obvious that a more efficient speaker can spell the difference between success and failure. Particularly demanding of amplifier power are small loudspeakers, a fair number of which fall at or near the lower end of our efficiency scale.
My original examples were chosen to cover the range of efficiencies in commonly available loudspeakers. The left column had figures for a speaker of 94 dB/W/m, but has been recalculated for Stuart's big Tannoys*; large horn-loaded systems (like Klipschorns) which are more efficient by an additional 5 dB, and all such speakers can produce very high sound levels with a modest amplifier.
While the table shows the effects of two of the three principal factors -- speaker efficiency and desired loudness -- it makes no mention of the listening room. The larger the room, the more power it takes to fill it, and the furnishings make even more difference than the size. These calculations are based on a room with a volume of just over 3600 cubic feet (17' x 25' x 8 ½') that is fairly "dead" acoustically, containing an area rug, heavy curtains, and several pieces of heavy stuffed furniture. This is a somewhat larger than average room; if yours is only half this volume, your power requirements will be less by about one third.
Increasing or decreasing the amount and type of furnishings will have a stronger effect, changing the power requirements over a range of about three to one. Our assumed listening room is fairly absorptive, which means that reflected sound is soaked up before it has a chance to contribute to the overall acoustic energy. A more "live" room, i.e. one with more reflective surfaces and less padded furniture, makes the amplifier's job easier.
The Digital Factor
People have often said that CD's demand more of our amplifiers than LPs do. This is true, but it doesn't affect your use of the figures in the table. Since the table gives the actual levels of acoustic events, an uncompressed CD that was recorded from a single microphone perspective will behave according to the numbers you see here. Older analog sources, or for that matter CD's on which accent mikes or compressors have been used, will have less dynamic range and so will be less demanding of amplifier power than the table predicts.
The level labelled "Cannon Fire (peak)" is approximately correct for a 105-mm Howitzer firing blanks at a distance of about 100 yards, as called for in the score of the Tchaikovsky "1812 Overture". The Telarc CD of this work contains cannon fire recorded at nearly full level. To reproduce this material at concert volume requires fully horn-loaded speakers of over 100 dB/W/m efficiency, but it can be done. As is obvious from the table, no amplifier is big enough to do it with a normal hi-fi speaker, and if it were the speaker wouldn't survive the attempt.
Note: Beware of the Telarc 1812; it contains lots of energy all the way down to 8 Hz, so although virtually no system is capable of reproducing sounds that low, all will be stressed by huge woofer cone motions if you play these cannon shots really loud. Ported speakers with no upstream infrasonic filter are especially vulnerable.
The Hidden Variable - Output Current
Until now we've been talking as though amplifier power ratings can be taken at face value. But not all amplifiers will deliver their rated power into a real loudspeaker. The figure on the spec sheet is measured with a pair of 8-ohm test resistors connected to the amplifier in place of speakers. But most "8-ohm" speaker systems have an impedance which falls to 5 ohms or less at some frequencies, causing them to draw more current than the resistor would.
In addition, the loudspeaker will present a reactive load to the amp that can increase the current demand still further, to as much as several times that demanded by a simple resistor. Since it is the current -- not the voltage -- that actually moves the speaker cone, output current capability is the limiting factor in most amps. Check the spec sheet, or published test reports, for the behavior of the amplifier at 4 and 2 ohms; its output should increase substantially over the 8-ohm value. Then test the amp with your own speakers -- or those you intend to buy -- before you commit yourself.
Finally, remember that you shouldn't be afraid of buying a bigger amp if you can afford it, especially if you like the features that come along with the extra power. That power will be used primarily on very brief musical peaks, and even a modest speaker system can safely absorb several hundred watts for a few milliseconds with no damage. It may actually be riskier to use an underpowered amplifier or receiver than an overpowered one; some protection circuits, when tripped, can produce ultrasonic oscillation at full power for long enough to burn out a tweeter.
This article and accompanying graph can be downloaded as a word processor file in two formats:
Word 6.0/95 [871Kb] and Word '97 [198Kb]
Text © E. Brad Meyer; HTML © SMR Home Theatre and Images © SMR Home Theatre, Seth Dominick (Judas Priest), La Scala Opera House & Telarc cannot be reproduced without permission.
* I am personally using a pair of Tannoy loudspeakers, and these are what E. Brad refers to above.
Power Rating: 120 Watt RMS, 500 Watt Peak. Sensitivity for 2.83 volt at 1 metre: 99 dB.
Nominal Impedance: 8 ohms. Frequency Response: 18Hz to 22kHz
Driver Type: 15-inch Dual Concentric. Enclosure Type: Compound horn. Dimensions: 1395 x 980 x 560mm.
Rosehall Industrial Estate,
Telephone: +44 (0) 1236 420199. Facsimile: +44 (0) 1236 428230
Return to the SMR Home Theatre menu