Amplifiers explained: a century of designs, and what the numbers mean

What an amplifier actually does

A source — a streamer, a turntable's phono stage, a CD player — produces a tiny voltage that carries the shape of the music. That signal has almost no power behind it: it can't move anything. An amplifier takes power from the wall and builds a much larger copy of that small waveform, with enough current to physically shove a speaker cone back and forth against air.

So there are two jobs, and they're worth separating: voltage gain (making the signal bigger) and current delivery (having the muscle to sustain it into a real, misbehaving load). The first is easy and solved. The second is where amplifiers differ, where the power supply matters, and where most of the price goes.

The ideal amplifier would be a perfect "straight wire with gain": identical output shape, no added noise or distortion, and total indifference to what's plugged into it. Every real design is a set of compromises against that ideal.

A short history: from one watt to a thousand

The triode era (1900s–1930s). Electronic amplification became possible with the triode valve, patented in the early 1900s. Early amplifiers made barely a watt or two, which is why the loudspeakers of the age were horns: horn loading was the only way to get usable volume out of almost no power. That constraint — efficiency first — shaped speaker design for decades.

Cinema and push-pull (1930s–1940s). Talking pictures needed real output for large rooms, and the push-pull output stage — two devices sharing the waveform — brought more power and less distortion. The 1947 publication of the Williamson circuit, a valve design built around negative feedback, set a new benchmark for low distortion and was copied and adapted worldwide.

The valve golden age (1950s). This is when domestic hi-fi as we know it appears, and several names in this catalog were at the centre of it. Quad's valve amplifiers became a British institution; Leak named a product the Point One after its distortion figure of 0.1%, a boast at the time; McIntosh built its reputation on a patented output circuit and its big power meters; and Marantz, under Saul Marantz, produced preamplifiers and monoblocks that still change hands for serious money.

Transistors take over (1960s–1970s). Silicon transistors were cheaper, cooler, more reliable and far more powerful per unit cost. Valves were pushed out of the mainstream within about a decade. But early solid-state amplifiers also earned a bad reputation — thin, hard, brittle-sounding — which is where the durable folk belief that "transistors sound harsh" comes from. It was a young technology being used badly, not a law of physics.

The specification wars (1970s). Two races ran at once: watts, and total harmonic distortion. Manufacturers piled on global negative feedback to drive THD figures down to absurdly small numbers on a test sine wave — and some of those amplifiers sounded worse than their specs implied, because a steady tone is not music. Research in that decade on transient intermodulation distortion made the point formally, and the industry slowly stopped treating a single THD figure as the score. Meanwhile designers went the other way for elegance: Quad's current-dumping topology of the mid-1970s is a famous example of solving the problem by circuit invention rather than brute feedback.

MOSFETs, Class A and the boutique era (1980s). Power MOSFETs arrived, and with them a high-end movement that deliberately traded efficiency for simplicity: low-feedback and pure Class A designs that ran hot and made modest power, sold on sound quality rather than numbers. Valves came back at the same time, no longer as the mainstream but as a deliberate choice.

Class D grows up (1990s–2000s). Switching amplification was an old idea that finally became practical for audio. Early implementations were used for subwoofers and car audio and were not taken seriously for full-range hi-fi. That changed: Bang & Olufsen developed the ICEpower switching platform that it still uses in its own active speakers and licenses widely.

Where we are now (2010s–today). Modern Class D modules measure as well as almost anything ever built, and they're small, cool and cheap enough to put inside the speaker. That's the real shift of the last decade: amplification is migrating into the cabinet — see Active vs passive speakers. In this catalog you can see it directly: Zu Audio's flagship carries an internal switching amplifier for its powered bass section, and Marantz's Horizon speakers use a modern GaN-transistor Class D stage.

Form factors: how the boxes are divided

Every amplifier chain does the same two things — select and control the signal (preamplification), then power it (power amplification). The form factors are just different ways of packaging those two.

• Integrated amplifier — preamp and power amp in one chassis, with input selection and a volume control. One box, one power cable, one purchase. For the large majority of systems this is the right answer, and it's what most people should start from.
• Pre / power separates — two chassis. The argument is isolation: the preamp handles millivolt signals, the power amp handles big currents and a large transformer, and keeping them apart reduces interference. It also splits the upgrade path in two.
• Monoblocks — one complete power amplifier per channel. Each has its own power supply and chassis, so the two channels can't interact at all, and you can place each one right next to its speaker and run very short speaker cables. Genuine engineering benefits, at double the cost, heat and shelf space.
• Stereo power amplifier — two channels in one chassis, fed by a separate preamp. The practical middle ground between separates and monoblocks.
• Receiver — an integrated amplifier with a radio tuner built in; the classic 1970s living-room box. An AV receiver extends that to many channels plus HDMI switching and surround processing, which makes it a different kind of product with different priorities.
• All-in-one / streaming amplifier — an integrated amp that also contains a DAC, network streaming and often a phono stage. One box replaces four. The trade-off is that the digital parts age faster than the analogue ones, so the whole unit dates sooner.
• The amplifier inside the speaker — in an active design there's no separate amplifier at all: each driver has its own amp channel in the cabinet. This is now the standard in studios and increasingly common at home.

One practical note on separates: an extra box means an extra pair of interconnects and another power cord, and a mismatch in gain or input impedance between a preamp and a power amp from different makers can cause real problems. Separates are not automatically better than a good integrated at the same total price.

Several brands in this catalog build amplifiers as their main business, or did historically — McIntosh, Marantz, Quad, Leak, Rega, Arcam, Cambridge Audio, PS Audio, T+A and Technics among them.

Operating classes: A, AB, D and the rest

The "class" describes how much of the time the output devices are conducting current. It's an engineering choice with real audible and practical consequences — and note that it is not a quality ranking. A is not better than D by virtue of the letter.

• Class A — the output devices conduct all the time, over the whole waveform. Inherently the most linear, with no handover point to get wrong. The price is brutal efficiency: typically only 20–25% of the power drawn becomes sound, the rest becomes heat, so Class A amps run hot, are heavy and make modest power for their size and cost.
• Class B — each device handles half the waveform and shuts off for the other half. Efficient, but there's a handover point where one device takes over from the other, and distortion right at that crossing is audible. Rarely used pure in hi-fi.
• Class AB — the practical compromise that dominated hi-fi for fifty years. The devices are biased to stay in Class A for small signals (where most listening happens) and slide into Class B for loud passages. Good efficiency, well-behaved handover, sensible heat.
• Class D — the output devices switch fully on and off at a high frequency, and a filter reconstructs the audio. Efficiency above 90%, so almost no heat, tiny boards and low weight. Note it is not "digital amplification" — the signal path is analogue, the devices just switch. Modern implementations are genuinely state of the art on the bench.
• Classes G and H — Class AB variants that switch or slide the power supply rails up and down to follow the signal, recovering some efficiency without Class D's switching. Common in high-power and pro amplifiers.

Valves: a device, not a class

Valve (tube) amplifiers are a separate axis: a valve amp can be Class A or Class AB, single-ended or push-pull. What makes them behave differently isn't the class but the device and the output transformer they need to drive a speaker.

• Higher output impedance. A valve amp grips the woofer less tightly than a solid-state amp does. That is the actual mechanism behind the familiar description of "looser, warmer" valve bass — it's an interaction with the speaker, not magic.
• Benign distortion character. Valves tend to distort in a way rich in even-order harmonics, which the ear finds pleasant rather than grating, and they clip more gently than transistors do.
• Impedance taps instead of one output. Many valve amps offer 4, 8 and 16 ohm connections, because the output transformer has to be matched to the load rather than simply coping with it.
• Single-ended triode (SET) designs make very little power — often 3 to 15 watts. They only work with genuinely efficient speakers, and they want a load that doesn't swing wildly.
• Maintenance is real. Valves wear out and need replacing every few thousand hours, and biasing may need occasional attention.

If a low-power valve amplifier is the goal, the speaker choice comes first, not second. In this catalog the natural partners are the high-sensitivity horn and full-range designs — Klipsch's heritage models, Zu Audio's crossover-less speakers, the horn-loaded Pylon Amber and Jade, and Fyne Audio's larger Vintage models. Several of these also present unusually benign impedance: Zu's DW6 and DWX are rated 12 ohms nominal and don't drop below 8.

Watts: the most misread number in audio

A power rating is meaningless without the load it was measured into. "100 watts" should always read "100 watts into 8 ohms", and ideally state that both channels were driven, across the full audio band, continuously.

The doubling test. Halving the impedance doubles the current demand. An amplifier with a stiff power supply rated 100 W into 8 ohms will deliver something close to 200 W into 4 ohms. One with a weak supply will manage 120 W and start to struggle. That single comparison tells you more about an amplifier's real capability than its headline figure does — and many manufacturers quietly omit the 4-ohm number.

Loudness is logarithmic, and this is the fact that changes decisions. Every extra 3 dB of output needs double the power. Roughly 10 dB is needed before most listeners call something "twice as loud" — and 10 dB means ten times the power. So going from a 50 W amplifier to a 100 W one buys 3 dB. It is a real but small change, and nothing like the difference the price suggests.

• Continuous (RMS) power measured across the band with both channels driven is the figure to trust.
• "Peak", "dynamic", "music power" and especially "PMPO" are marketing figures. PMPO in particular can be an order of magnitude above anything the amplifier can actually sustain.
• Headroom is why you want more than the minimum. Music has short peaks far above its average level; an amplifier running out of room clips them.
• Too little power damages speakers more often than too much. A small amp driven into clipping sends a distorted, squared-off waveform with abnormal high-frequency energy straight at the tweeter. Tweeters die of clipping far more than of raw watts.

Every speaker page on TrueScale lists the maker's recommended amplifier range. A good rule is to aim for the upper half of that range from a competent amplifier, rather than the maximum from a weak one.

Sensitivity beats watts — by a factor of 800

How loud your system plays is decided by the speaker at least as much as by the amplifier, because sensitivity sets how much sound you get per watt. And the spread across real speakers is enormous.

Real speakers span an enormous range: from about 78 dB (the Radiant Acoustics Clarity 4.2, a small, deliberately low-sensitivity monitor) up to about 107 dB (the Klipsch Jubilee, a large horn). That is a 29 dB gap — which, at 3 dB per doubling, means the Klipsch needs roughly one eight-hundredth of the power for the same volume. One watt on the horn does what eight hundred watts would do on the little monitor.

That is the whole reason an eight-watt valve amplifier can be a serious choice, and the reason a 300-watt monster can still sound strained on the wrong speaker. Sensitivity and impedance together decide what amplifier you need — the full treatment is in Sensitivity and impedance.

Impedance: the number that lies

A speaker's impedance is not a fixed value — it's a curve that rises and falls across the frequency range, driven by the drivers' motors, the cabinet tuning and the crossover. The "8 ohms" or "4 ohms" on the box is a nominal average, and the amplifier doesn't see the average. It sees the dips.

How much does that matter? Plenty of speakers sold as 8-ohm drop to 4.5 ohms or below. The Focal Aria Evo X N°4 is nominally 8 ohms and dips to 2.5. At the extreme end, MartinLogan's electrostatic hybrids fall below 1 ohm at the top of the audio band — the Neolith reaches 0.43 ohms, roughly a tenth of its nominal rating.

A low dip means a big current demand, and an amplifier that can't supply it will lose control, compress, distort, or trip its protection. Which is why speakers like the MartinLogan Masterpiece electrostatics genuinely require an amplifier that is happy driving 2 ohms and below, not merely one with a big wattage sticker.

Two related ideas worth knowing:

• Output impedance and damping factor. A solid-state amplifier has a very low output impedance, which means it electrically brakes the woofer's motion — this is what "grip" or "control" in the bass refers to. Valve amplifiers have much higher output impedance and therefore less of it. Once damping factor is high enough, more is not audibly better; it's a threshold, not a score.
• 4 ohms is no longer the exception. A modern catalog carries about as many 4-ohm speakers as 8-ohm ones, so the old assumption that a speaker is "probably 8 ohms" is simply out of date, and an amplifier that is only comfortable at 8 ohms rules out a large part of the market.

Which specifications actually predict anything

Worth reading:

• Continuous power into the impedance you will actually use, both channels driven.
• Whether power roughly doubles from 8 to 4 ohms — the power supply's report card.
• Stability into low impedance (a stated 2-ohm rating, or the maker being explicit about difficult loads).
• Input sensitivity and gain, if you're mixing a preamp and power amp from different brands.

Mostly noise:

• THD figures below roughly 0.1%. Below that level the number stopped correlating with what people hear, which is precisely the lesson of the 1970s specification race described above. A lower figure is not a worse one — it just isn't informative on its own.
• Frequency response extending to 100 kHz. Fine as evidence of a wide-bandwidth design, irrelevant as something you will hear.
• PMPO and peak wattage. See above.
• Enormous damping factor claims. Past a threshold, marketing.

Matching an amplifier to your speakers

A practical order of operations:

• Start from the speaker, not the amplifier. Read its sensitivity, its nominal and minimum impedance, and the maker's recommended power range — all of which are on every TrueScale speaker page.
• Aim for the upper half of the recommended range. More clean power than you need is cheap insurance; too little is the actual risk.
• Check the low-impedance behaviour if the speaker dips. A 4-ohm nominal tower with a 2.5-ohm minimum needs an amplifier that welcomes it.
• For valves, choose the speaker first — around 90 dB sensitivity as a floor for a low-power amp, and a reasonably flat impedance curve.
• Account for the room and how loud you listen. A big room and high levels multiply the power requirement; a small room at moderate volume needs remarkably little.
• Consider skipping the amplifier entirely. If the appeal of a matched, finished system outweighs choosing components, an active speaker already contains amplification designed for its own drivers.

The bottom line

A century of amplifier design has been one long negotiation between three things: linearity, efficiency and cost. Valves and Class A bought linearity with heat and watts; transistors and Class AB bought power and reliability with complexity; Class D bought efficiency and made it small enough to hide inside a speaker. All four are alive today because each still wins on some axis.

What matters when you buy is narrower than the discourse suggests: enough clean power for your speaker's sensitivity, stability into its real impedance, and a format that fits how you want to live with the system. Everything after that is preference — which is a legitimate reason to choose, just not the same thing as a specification.

Every speaker page here lists the three numbers that drive the decision — sensitivity, impedance and recommended power. You can put four spec sheets side by side in TrueSpecs to compare how demanding your shortlist really is, or read How to read a speaker spec sheet for the rest of the numbers.