Sweet spot of amplifier power

For quite some time, and according to the audio mainstream, relation between amplifiers power and their supposed quality looked like relation of equivalence. Most amplifiers manufacturers were setting their lines of products in a way that tightly associated power to the quality, which of course also implied sound quality. And to certain degree, it is still like that.

Supposedly, power is good even if it is nominally not needed, because “power reserve makes amplifiers sounding better”, and if nothing else it is welcome as just in case feature (“better to have it and not need it, than the opposite”). Finally, today’s ubiquitous D class (“digital”) amplifiers, due to their high efficiency and consequently high power, are easily promoted and understood by wider audience as a progress.

Yet, as opposed to the said mainstream, hi-end audio devotees are not that obsessed with high power amplifiers anymore. Low Wattage systems are more than legitimate part of modern hi-end audio world. So, telling today how high power is not necessary is nothing new or brave, as it was back in 70s, when Jean Hiraga came with 8 W class A “Monster” amp. However, real technical reasons behind apparent move to the lower power amps are not spelled out very often. And there are two major reasons why high power can be not only unnecessary, but also unwanted.

The first is about active devices. Lower power active devices performance is superior to that of higher power ones. This applies to both transistors and tubes. Using several active devices in parallel to increase the power doesn’t help quality, either, it only brings more problems to solve.

And second, practically every active device has certain operating conditions (supply voltage and bias current) that set sweet spot for its best sound. In turn, supply voltage is directly associated to the available power, and best sounding voltage is usually way below commonly regarded safe operating maximum.

And in addition to the inherent amplifier properties, there are several other reasons why the best sound is not associated to high power, such as previously explained superiority of lower power mains transformers.

Just to make things clear: I am far from advocating the lowest possible power. But “the more the better” argument obviously doesn’t hold either. Simply, for good power amp design, the power itself is of secondary importance.

That is about the amplifier itself. Of course, amplifier must be still able to drive loudspeakers. But, what the mess about enormous amplifier power in home audio is actually for? Amplifier is not supposed to move the car, but loudspeaker cone. Usually it is a few tens of grams, and it has to be moved for a few millimeters forth and back. Hundred grams and a few centimeters would be the “worst case”. It really is not a tough work. Still, loudspeakers efficiency is quite poor, and typical 88 – 91 dB / 1 W / 1 m sensitivity equals to 1 – 2 % overall power efficiency. And some loudspeakers have even lower efficiency. Some have low impedance, which demands even more current, and non-linear impedance phase is sometimes discussed as another  part that can also increase actual current requirements.

So, how much power is actually enough? Let’s recall some classic rules about loudspeakers sensitivity and loudspeakers impedance.*

With loudspeakers sensitivity of 87 dB / 1 W / 1 m, a 20 W (that is 13 dBW i.e. 13 dB above 1 Watt) will produce 100 dB sound pressure, which is nominal equivalent of jack hammer. In my experience, in usual home conditions it is indeed enough. Of course, change in speakers sensitivity for 3 dB will double or halve requirements in amplifier power, so 93 dB sensitive speakers will take only 5 W for the same 100 dB sound pressure.

“Power” is however still made of voltage and current, and loudspeaker’s impedance will tell their mutual relation, or in other words, it will tell loudspeakers actual voltage and current requirements.** And simple power equation (P = U x I) and good old Ohm’s law (I = U / R) might solve most of your doubts here. Also, by combining these two equations, once you know the speaker impedance, you can express the power solely by either voltage (P = V2 / R) or current ( P = I2 x R). As previously said, non linear impedance phase can increase demands in current, but this increase is usually not higher than 20-30 %, and if impedance minimum is associated to the voice coil DC resistance, which usually is the case, and at which point the impedance phase is actually linear, then impedance phase can be mostly neglected.

The bottom line here is that conventional loudspeakers with nominal impedances above 6 Ohm don’t demand that much of current. At 8 Ohm, 2 – 3 Amperes equals 30 – 70 W, and it is rather voltage that sets the limit in this case. However, when speaker impedance gets low, it is important to know that more Watts at typical 8 Ohm load don’t help – it takes more current instead.

So, every speaker definitely has its own demands from amplifiers. But amplifiers also have the logic of their own. And whether you start with amplifier and look for speakers to match, or the other way round, the “trick” is to let the amplifier operate the best way, with speakers it is able to drive, both in terms of voltage and current.



* – As a side notice, let me say that traditional way of determining required amplifier power per loudspeakers power is mostly pointless. Loudspeakers actually have no power, they are passive components that have only power handling (capacity). It is OK if you consider this handling too, so you don’t fry your precious loudspeakers too easily, but apart from that, such an approach is in fact useless.

** – Actually, there is a paradox about loudspeakers operation here. Loudspeakers are normally designed to work with voltage sources, whereas their cone move is actually proportional to the applied current. One might think about possible current-drive (counter)revolutions here, but I believe voltage requirement is the fact to live with – and so long as we are in the voltage world, we should think the current as a consequence of voltage, associated to the actual speaker impedance.



Related topic:

Model A, one year later: Power