How Much Power Do You Really Need?

What’s the Number One question demanded of self-styled audio experts like me? “How much power do I need?”

But expert or not, I can’t answer it: not without knowing the details of your speakers, your room’s size, layout, and furnishings, your listening habits, musical tastes, boxers-or-briefs, and a dozen other details. I can, however, arm you with a few concepts to think about, concepts that, if mastered, hold the key to all power. Well—audio power, anyway.

What's In a Watt?
First, foremost, and forever, the diabolical decibel (dB). To hold a meaningful conversation we must talk amplifier power in terms of decibels, a scale that expresses power comparisons the way we hear: logarithmically. A decade or so back, a movement to make the “dB watt”—decibels in reference to 1 watt—a spec-sheet staple enjoyed brief traction, but today the abbreviation “dBw” is rarely seen. Yet it’s a superbly useful and far more meaningful way to describe amplifier capabilities—and once you get the basics, you can roughly convert spec-sheet wattages to dB watts easily enough. The dB watt’s key value is this: Assuming reasonably linear loudspeaker behavior, dBw translate more or less one-for-one to the increase in useful loudness you can expect from a given loudspeaker in a given setup.

We call an output of 1 watt, connected to a loudspeaker-like test-load of 8 ohms (effectively 2.83 volts from the amplifier,) “zero dB-watts,” or 0 dBw. That is, relative to 1 watt, 1 watt is precisely zero greater or smaller (duh!). (Remember, all decibel expressions are always at their core expressing a ratio: comparisons, usually to a standardized reference. In the dBw case the standard is 1 watt.) For mathematical reasons I ain’t going into here, 2 watts equals 3 dBw, and 4 watts 6 dBw. Ten watts is 10 dBw, and 100 watts is 20 dBw. Mercifully, logarithms simply multiply their ratios, so 30 dBw, for example, equals 1,000 watts (10 dB more than 100 watts or 10x20 dBw, which is 10x100 watts.) Stay with me now, here’s the nub: this means that doubling amplifier power yields only a 3 dB gain in “how-loud-ness,” while increasing power ten-fold yields a mere 10 dB increase.

Great! What do “only” and “mere” mean? In real-world terms, a 3 dB loudness increment is quite small—in fact, it’s the smallest change in loudness most untrained listeners will call noticeable. A 10 dB change is characterized by the general populace as “twice as loud,” whatever that means. (That’s like saying that Emilia Clarke is “twice as hot” as Mila Kunis.” But I digress.) That’s right: you must double power to effect any meaningful difference at all in perceived volume, and you must times-ten it to make things really, solidly, usefully louder. All this presupposes one amplifier channel driving one loudspeaker, just to keep it simple. (If we’re talking stereo, with two identically powered channels driving two identical speakers, we get about 3 dB more loudness (or dynamic potential) off the baseline thanks to the laws of acoustics).

Let's Be Sensitive
Our next factor is speaker sensitivity. Two different loudspeakers served identical 1-watt signals will play louder or softer depending on their sensitivity (sometimes referred to as "efficiency," though that term is less technically correct in this instance). A typical consumer speaker given 1 watt and measured via a microphone placed 1 meter away will produce about 87 dB sound pressure level or SPL (“how loud,” in decibel terms). Choose instead a speaker of 90 dB SPL/1 watt/1 meter sensitivity and you just made the equivalent change in perceived volume as doubling your amplifier power without increasing your actual amp power by so much as a milliwatt, or your amplifier budget by so much as a dime.

Speaker sensitivity has nothing to do with speaker quality, sonic or otherwise. But if you’re torn between two similar-price speakers, choose the one with higher sensitivity.

Let me underscore that speaker sensitivity has nothing whatsoever to do with speaker quality, sonic or otherwise; it’s a simple quantification of how-loud-per-watt. But if you’re torn between two similar-price speaker models you love equally, choose the higher-sensitivity one. One caveat, and it’s a biggie: due to measurement techniques, environments, and plain old manufacturer mendacity, spec-sheet sensitivity ratings can be extraordinarily variable, so take claimed differences with a double shake of salt. You can check the Test Bench box in our speaker reviews for a statement of each loudspeaker model’s sensitivity that is consistently derived, and easy to compare, across all products.

Then there’s distance. What happens if we listen from a distance of, say, 2 meters instead of 1? The sound is quieter (duh!), but by how much? In “free space” (outdoors) sound pressure falls of at the inverse of the distance—the sound is 3 dB softer for a doubling of listening-distance, or 6 dB softer for quadrupling it. When indoors things are more complicated, but if you figure a total of about 4.5 dB to 6 dB softer for each doubling of the listener-to-loudspeaker dimension you won’t be too far off. Obviously, then, bigger rooms with more distant seating positions will require more amplifier power, which is why stadiums require many kilowatts while headphone amps, with a listening-distance measured in millimeters, manage with milliwatts (thousandths of a watt).

Dynamic Amp Personalities
Lastly, there’s dynamics. All of these measures of power in watts and dB/SPL are derived using steady-state test signals that have effectively zero dynamic range: a peak-to-average ratio of 1. With real program material, the more dynamic kinds of music have a peak-to-average ratio of 20:1 or even higher. This means that if 2 watts per channel steady-state produces a comfortable musical average level from a given loudspeaker—and it very well might—something north of 200 watts, even if for only a few milliseconds at a time, will be required to produce peak musical transients (without any limiting or distortion) at something approaching concert levels. And if the loudspeakers are a little low in sensitivity, or the room is extra large, or “concert level” means 120 dB/SPL rock-concert level, this second number could easily crest 500 or even 1,000 watts, or, with low sensitivity speakers in a large room, even 10,000. Per channel. And that’s assuming the speaker could withstand that much input power, even briefly, without bursting into flames.

Put it all together, and you begin to see that replacing that aged 5x75-watt receiver with one rated for 125 watts per channel is ineffectual, if not downright foolish in terms of achieving a noticeably higher volume level. Moving your sofa a couple feet closer to the speakers might be just as useful, and it costs nothing.

Not to say that power in and of itself is a bad thing; in fact, more power is always better than less power. (Clean power, responsibly used, rarely damages loudspeakers, while underpowered amps driven deep into long-duration clipping frequently do.) Of course, this cuts both ways. If you listen just “a little bit” softer, you can halve your power requirements (our 3 dB rule in reverse); listen a little bit softer still, and you can quarter them.

Just the same, keeping any amplifier in its linear operating range, completely free of distortion, even for any small fractional second, should always be job one. And in my view this—not silver wires or gigaHz resampling rates—is the principal factor that makes “high-end” systems sound high end.

So on reflection, it seems I actually can answer the “how much power” question I posed at the top: More!

COMMENTS
drewdlz's picture

This is one of the better articles explaining this that I've read. Wish I learned this sooner than I did.

One question about the end of the article where you say, "keeping any amplifier in its linear operating range..." How is it that you can know if you are doing that or not? Sorry if that's a dumb or obvious question.

hk2000's picture

In my experience, most amps/receivers dont start distorting until well beyond the midpoint- especially those from reputable manufacturers (Marantz, Denon, Onkyo, ATI, Bryston, Parasound,..etc.)Your best bet is to read reviews and check out the measurements. S&V is one of the best in that regard- their graphs usually show you exactly when the amp STARTS to distort, so you can know what the limit is for very clean sound. If you know your speakers sensitivity, you will know what level not to exceed.
For example, if your speaker has a 85db sensitivity, you need one watt for 86db, 2 watts for 89 db, 4 watts for 92 db 8 for 95, 16 for 98, 32 for 101, 64 for 104 and so on, so a 100 watt receiver will give you 105 to 106db of maximum clean sound.

hk2000's picture

*I meant 86 not 85db

normie100's picture

No, you meant 85dB. You have to add the dBs, and 1W is 0dBW. When added to the sensitivity of your speaker of 85dB/W/m, you get 0+85=85dB at one meter. Then 2W is 3dBW above this as 2W is double 1W, for 88dB. And 100W, being 20dBW, gives you 105dB.

dommyluc's picture

I figure that, if I have my windows open, and the neighbors can hear me playing Steely Dan or Debussy a block down the road, I have enough power.
Oh, and I make $25,000 a week just by being my cute and cuddly self. Find out how!

http://www.Hey, S&V! How about getting rid of these stupid trolls who post their useless links and take over the comments section?
http://www.Hey, S&V! How about getting rid of these stupid trolls who post their useless links and take over the comments section?
http://www.Hey, S&V! How about getting rid of these stupid trolls who post their useless links and take over the comments section?
http://www.Hey, S&V! How about getting rid of these stupid trolls who post their useless links and take over the comments section?

brenro's picture

A speaker's impedance over its operating range also is a big factor in how much amplifier power you need. 4 ohm or lower speaker impedance requires an amplifier to work much harder.

normie100's picture

The article states that in free space where there are no reflections, a speaker's apparent loudness will halve (decrease by 3dB) for every doubling of distance. This means, for example, if 104dB at 1m, then it will be 101dB at 2m, 98dB at 4m, and 95dB at 8m.

Most engineering papers (sites) indicate that the value is actually 6dB per doubling due to the Inverse Square Law (http://www.engineeringtoolbox.com/inverse-square-law-d_890.html). Conversely, in the following paper (http://www.linkwitzlab.com/rooms.htm), a more lively room will contain 3dB more SPL than a dead room (dependent on frequency, volume, etc.).

X