Power Plays: Making Sense of Power Specifications for Amplifiers, Receivers, Speakers and Subwoofers
By Michael Riggs (10/18/00)

Of all the many performance specifications people are confronted with in audio, the one they are perhaps most comfortable with is power. Unlike, say, signal-to-noise ratio or sampling rate, power is something that comes up routinely in everyday life. How many watts is that light bulb? If it's a 100-watt bulb, it will be brighter than if it is just a 60-watter. More is better, right?   Unfortunately, the situation is a little more complicated in audio, where the question "How many watts?" can have several different answers, each perfectly legitimate in its own way. Still, some answers are better than others, so it's useful to know how to tell the watts from the watts.

Amplifiers and Receivers
Amplifier power ratings are important because amplifiers do the work of making loudspeakers produce sound. (Receivers incorporate amplifiers, so everything said here about amps applies to them as well.) And the more power something has, the more work it can do. So all else being equal, the more watts available the better. A more powerful amplifier should be able to play louder without distortion than a less powerful one. Here's where it starts to get tricky. Sometimes you may hear or read someone saying that a particular amplifier sounds more powerful than one with a higher power specification. Sometimes that's just evidence of an overactive imagination at work, but such things really do happen. There are two main reasons, one having to do with how manufacturers specify power, the other with what is involved in delivering power to a loudspeaker.   Rating games. More than 20 years ago, the Federal Trade Commission (FTC) laid down rules for how power specifications for home audio amplifiers could be advertised. Basically, the rules require manufacturers to declare how much average (as opposed to peak) power an amplifier can deliver into a stated impedance over a stated bandwidth with a stated maximum level of distortion and with all channels driven simultaneously. They also require that the primary specification be for an 8-ohm impedance. A manufacturer can list specs for other impedances, but they must be subsidiary to the 8-ohm rating. (We'll get into the significance of that restriction later when we discuss delivering power to speakers.) A typical stereo amplifier power specification might thus read something like the following: 100 watts per channel into 8 ohms from 20 Hz to 20 kHz with no more than 0.1% THD (total harmonic distortion), both channels driven.   Although one can quibble with some details of the FTC strictures, they are basically good in that they force amp makers to give the most important specification in a form that is readily comparable from brand to brand and model to model. There is still some wiggle room, however. For example, a manufacturer can usually report more power for his amplifier if he raises the distortion threshold or narrows the bandwidth. The same amplifier described in the last paragraph might also be (legitimately) rated as follows: 115 watts per channel into 8 ohms at 1 kHz with no more than 0.8% THD, both channels driven.   Another interesting dodge has emerged in the home theater era, based on the fact that so many amplifiers and almost all receivers now have five channels rather than just two. Yet many five-channel amps and, especially, receivers sport power ratings based primarily or exclusively on two-channel operation, or on some other combination short of all channels driven simultaneously. Reason: The fewer channels operating, the more power the ones that are running can deliver. So the manufacturer can say he's got, say, a 120-watt amplifier instead of a 100-watter. It's just a matter of less competition for the amplifier's power-supply resources. The argument for this method of specification is that it is more realistic, since it is very rare that all five channels would be required to deliver full power simultaneously anywhere except on a test bench. Unfortunately, it also can be used to paper over design shortcomings. And since there is not much uniformity in how manufacturers apply this sort of rating method, it makes comparisons more difficult.   At Harman, we think the fairest, most honest, and most revealing rating method is the obvious one: to specify power with all channels fully driven. And that's how we do it for our products.   Delivering power to speakers. The other thing that complicates making sense of amplifier power specs is that loudspeakers are complex loads. For simplicity and ease of comparison, amplifier power measurements are made into purely resistive loads, usually 8 ohms. But speakers, almost without exception, present complex impedances. That means their impedances contain reactive (inductive and capacitive) elements as well as resistive ones, which in turn means that their impedances vary with frequency. A speaker having a nominal impedance rating of 8 ohms will have that actual impedance at some frequency or frequencies, but you might reasonably expect it to go down to 5 or 6 ohms (or below) at others and much higher at still others. The rated impedance is thus an approximation, at best.   Why does that matter? Let's consider what a watt really is. Power (watts) is voltage (volts) times current (amperes, or amps): 1 volt x 1 amp = 1 watt. An audio amplifier normally is designed to approximate what is known technically as a "constant-voltage" device, meaning that a given voltage in will yield a given voltage out, determined by the amplifier's gain. Put 1 volt into an amplifier with a gain of 10, and you should get 10 volts out. How much current you get out is then determined by the impedance of the device the amplifier is driving. Assuming a perfect amplifier, current out should equal voltage out divided by the load impedance. Consequently, halving the impedance (going from 8 to 4 ohms, for example) should double the current. And since power is voltage times current, that means the power should double as well.   With real amplifiers, however, it usually won't. The main reason is that amplifier output transistors are limited in how much current they can transfer without overheating and destroying themselves. A designer can get around this problem by using more robust transistors, or simply more transistors, but that drives up cost. Any practical amplifier will therefore have some limit on the amount of current it can deliver, set primarily by the output transistors, just as it will have a limit on its output voltage, determined primarily by the design of the power supply (if not the AC line voltage). The design engineer has to decide what those limits will be, based on some cost/benefit analysis.   Because amplifier power tends to be specified mainly for 8-ohm loads, there is a commercial incentive to concentrate on the voltage side of the equation, because into 8 ohms that will normally be the limiting factor. Unfortunately, loudspeakers often present a lower impedance–somtimes a much lower one–right in the frequency range where much musical energy is concentrated. And for reasons too complicated to get into here, reactive impedances, such as those presented by loudspeakers, can often draw more current than the simple magnitude of the impedance would suggest. An amplifier designed with inadequate concern for current capacity may therefore fail to perform up to the promise of its power spec when it is required to drive loudspeakers instead of 8-ohm laboratory resistors.   You can guard against this problem by looking for a 4-ohm power specification–preferably one that is at least 50% above the 8-ohm spec. It is also a good (though not infallible) sign if the manufacturer makes a point of current capability in describing its amplifiers and receivers.   One last thing that can affect amplifier power specifications is simply how conservative the manufacturer is in rating its products. Is the goal to print the best possible spec or, perhaps, to ensure that the customer will get the rated output even if the AC line voltage at his home is only 110 volts rather than 120? This is hard to assess except by reputation and track record, but it is a significant factor.

Loudspeakers
Power specifications for speakers are a somewhat simpler matter. In the case of passive speakers (those without internal amplifiers), any power rating will be for how much the manufacturer thinks the speaker can take without damage. (Or, in some instances, the minimum the manufacturer thinks the amplifier should be able to provide in order to provide satisfactory listening levels.) Unfortunately, the amount of power a speaker can take is a complex function of spectral energy distribution (the relative proportion of energy versus frequency) and duration. So there really is no one definitive number. Worse, no universally accepted standard exists for determining speaker power handling, which means that two different manufacturers could come up with entirely different numbers. About all you can determine from speaker power-handling specs is whether one model in a company's line can take more punishment than another. You can't compare realistically between brands, and even within a brand, you can't take the absolute numbers too seriously.   This is true for powered speakers as well, though for different reasons. By powered speakers, we mean those that have their own internal amplification. Very few full-range speakers are fully powered, though there has been a trend in recent years building in woofer amplifiers. Most separate subwoofers, on the other hand, do incorporate their own amplifiers. When a speaker does include amplification, the manufacturer usually will give a power rating for that amplifier–which you should simply ignore.   If power is so important a specification for separate amplifiers and receivers, why isn't it for a powered speaker or subwoofer? Actually, it is important, but only to the designer. He's the one who has to figure out what combination of speaker sensitivity and amplifier power will yield the desired sound output capability, which is what really matters. What you need to know is how loud the speaker or subwoofer can play over what frequency range and with how much distortion. In other words, you need to know what's coming out of the speaker, not what's going into it. It is entirely possible, for example, that a subwoofer with a 100-watt amplifier could outperform one with a 200-watt amp in every respect. For a consumer, the power rating of an amplifier built into a speaker is an absolutely useless number.   In addition, there is the matter of the accuracy of such ratings. Because amplifiers built into speakers or subwoofers are inaccessible, it is rare that anybody checks how much power they can really deliver. Power numbers sell, which creates a temptation to–how shall we put this delicately–exaggerate that some companies find irresistable. Just one more reason to ignore them.

Final Important Note
True or false: A 100-watt amplifier will play twice as loud as a 50-watt amplifier. Answer: false. The 100-watt amp will play 3 decibels (dB) louder, which is noticeably but not extravagantly louder. Every doubling of power will give you another 3 dB. To achieve a doubling of perceived loudness you need approximately ten times the power (which works out to 10 dB). The reason is that our ears respond to changes in sound pressure logarithmically rather than linearly. This is why you see some many audio specifications made on the decibel scale, which also is logarithmic.   What this means to a shopper is that differences in amplifier power of less than a factor of two are pretty much inconsequential. If you're trying to choose between a 100-watt receiver and a 120-watt model, or even a 150-watt model, power should not be the deciding factor.   It also means that speaker sensitivity can be pretty significant. If a particular speaker has 3 dB greater sensitivity than another, it will require only half the amplifier power to produce any given amount of sound output. Speaker sensitivity is normally specified as the sound-pressure level produced at a distance of 1 meter for an input of 2.83 volts (which is equivalent to 1 watt into 8 ohms). You might see this printed in rather shorthand form, such as: 90 dB SPL/1 watt/1 meter. Any way it's written, it's a number well worth paying attention to.