Valve, solid-state or Class D: how to choose the right amplifier topology for your system

A customer walked into my old shop on Smith Street in Fitzroy carrying a pair of KEF LS50s and a freshly purchased 300B single-ended triode amplifier. Eight watts. I didn't have the heart to tell him immediately. We brewed tea, put on a record, and within about thirty seconds the penny dropped — the bass collapsed the moment the music got complex, and the midrange, gorgeous as it was, floated above nothing.

That moment is burned into my memory because it's the most instructive amplifier lesson I've ever witnessed. Not a measurement. Not a spec sheet. Just the sound of a beautiful circuit trying to do something it was never designed for.

The valve versus solid-state versus Class D question gets asked constantly, and most answers are either gear-head tribalism or marketing copy. So here's what I actually think, after years of living with all three in real systems.

Why topology matters before anything else

An amplifier's topology — the fundamental circuit architecture — determines how it generates power, how it behaves as speaker impedance changes, how it distorts when pushed, and how much heat it throws into your room. These aren't audiophile abstractions. They're practical variables that change whether a given amp will work with a given speaker in a given space.

The three broad families are: valve (tube) amplifiers, solid-state amplifiers (which covers Class A, Class AB, and Class B designs), and Class D switching amplifiers. They overlap in some applications and are genuinely incompatible in others. No topology is universally superior. Anyone telling you otherwise is selling something.

Valve amplifiers: the ritual and the reality

I'll admit my bias upfront. I own four valve amplifiers and I'm not selling any of them. But I've also watched enough people buy the wrong one to write this section with clear eyes.

Valve amps — particularly single-ended triode designs and lower-power push-pull designs — produce their characteristic distortion almost entirely in even harmonics. The second harmonic especially. Human hearing is quite tolerant of even-order harmonics; they relate to the natural overtone series of acoustic instruments in a way that odd-order harmonics (more common in clipped solid-state designs) don't. Whether that explains the warmth people hear or not is a separate debate. The effect is real regardless of the mechanism.

What's also real is the output transformer. Every valve amp uses one to match the high-impedance output of the tubes to the low-impedance load of a speaker. That transformer is simultaneously the amp's greatest engineering challenge and the source of much of its sound. A well-designed output transformer — from a maker like Audio Note, Line Magnetic, or Shindo — costs serious money and takes decades to get right. A poorly designed one rolls off the frequency extremes and smears transients. The transformer is usually where the money goes in a valve amp, and where corners get cut in cheap ones.

Power output from a valve amp is almost always lower than an equivalent-priced solid-state design. A single-ended 300B circuit gives you somewhere between 7 and 10 watts. A push-pull EL34 design might deliver 35–40 watts. An KT88-based push-pull could hit 50–80 watts. These numbers feel modest until you understand that valve watts are not the same experience as solid-state watts — partly because valve amps clip softly and gracefully rather than hard, so they don't sound strained at the same subjective level. Still, physics is physics. Eight watts will not drive a 4-ohm, 83dB speaker to realistic levels without audible compression.

The impedance question is where most people trip up. Valve amps typically have a higher output impedance than solid-state designs — sometimes 1–4 ohms, depending on the circuit and the output transformer tap you're using. Pair that with a speaker whose impedance swings wildly across the frequency range (some electrostatics drop to 1–2 ohms in the treble; some bass-reflex designs do something similar around tuning frequency) and the frequency response of the combination will not be flat. That's not automatically a problem — many people find the result pleasant — but it is a real interaction you need to understand, not ignore.

The practical upshot: valve amps reward high-sensitivity speakers. 90dB and above is a reasonable starting point. 93dB-plus gives you real headroom. Something like a high-efficiency horn speaker, a classic BBC-style standmount, or a well-designed single-driver design will often sing with a modest valve amp in a way that a 6-ohm nominal modern speaker simply won't. Check the KEF LS50 Meta (check price) specs — 85dB, 8-ohm nominal but dipping lower — and ask yourself honestly whether your 8-watt SET is going to cope.

Solid-state amplifiers: the workhorse case

Class A solid-state runs the output devices in conduction for the full audio cycle, which eliminates crossover distortion entirely. It runs hot, uses lots of power, and tends to produce amplifiers that are physically heavy and expensive to build well. Pass Labs, First Watt, and Sugden are the names that come up most often in this space. Class A solid-state is the closest topology gets to a valve amp in terms of distortion character — even-order, low-level, smooth — while retaining the low output impedance and high current delivery that drives difficult speakers without complaint.

Class AB is the dominant solid-state topology for anything above about 50 watts. It runs the output devices in Class A for small signals and transitions to push-pull operation as the signal grows. Done well, crossover distortion is pushed well below audibility. Done poorly, it's the source of that particular solid-state glare that gave transistor amps a bad reputation in the 1970s. Modern Class AB designs from serious makers are very good. The McIntosh MA352 (check price) — which pairs solid-state power with a valve input stage — is a useful example of how the line between topologies can blur in practice.

The key advantage of solid-state is the low output impedance, typically a fraction of an ohm, which means the amplifier's frequency response barely changes regardless of what the speaker's impedance does. A solid-state amp with a damping factor of 100 or more is essentially unaffected by the speaker's impedance curve. This makes solid-state the sensible choice for speakers with complex crossover networks, low sensitivity, or low impedance loads — which describes the majority of modern loudspeakers.

Power is cheaper in solid-state, too. A 100-watt Class AB amplifier costs a fraction of what a 100-watt valve amp would, and the 100 watts is genuinely usable across a wide range of speaker loads. If you're running something like the Focal Kanta No.2 (check price) — 91dB but with a complex impedance curve — solid-state is the more predictable partner.

Class D: not what it used to be

Class D gets unfairly dismissed in analogue-leaning circles. I'd be lying if I said I wasn't one of the dismissers for a long time. The older switching amplifier designs — and plenty of budget ones still — had a particular graininess in the upper midrange and treble that I found fatiguing on anything with significant harmonic content. Strings, vocals, acoustic guitar. It wasn't subtle.

But the technology has moved substantially in the last decade. Modern Class D designs from companies like Purifi, Hypex, and Mola Mola (which builds its own proprietary switching stage) measure extraordinarily well — lower distortion and noise than most Class AB designs — and the better implementations sound it. The efficiency advantage is real and meaningful: a 200-watt Class D amplifier draws a fraction of the power and generates a fraction of the heat of an equivalent Class AB design. For a streaming integrated in a living room, that matters.

The remaining caveat is that Class D amplifiers can interact oddly with certain speaker loads at high frequencies, particularly speakers with unusual impedance behaviour in the treble. The output filter in a Class D amplifier is designed for a specific load impedance; deviate significantly from that and the frequency response at the amp's output changes. In practice most speakers are fine, but it's worth checking if you're running something unusual.

For a useful reference point on how modern streaming amplifiers (many of which use Class D or hybrid Class D output stages) handle real-world system matching, our guide to the best streaming amplifiers covers a range of topologies in that context.

Power and impedance: the matching problem in plain terms

Sensitivity and impedance are the two speaker measurements that determine whether an amplifier will be comfortable in a system. Sensitivity tells you how loud the speaker gets from one watt at one metre. Impedance tells you how hard the speaker is to drive electrically.

A speaker rated at 88dB sensitivity and 8-ohm nominal impedance is moderately easy to drive. A speaker rated at 84dB at 4 ohms is significantly harder — it needs more current, and the lower impedance means the amplifier works harder for the same output level. Halving the impedance roughly doubles the current demand. Solid-state amplifiers typically double their power output as impedance halves (down to a point). Valve amplifiers, working through a transformer, do not behave this way — their output is more fixed, and some will actually lose power into very low impedance loads.

The rough rule I use: for every 3dB of sensitivity the speaker lacks relative to an easy load (say, 90dB), you need to double your amplifier power to achieve the same listening level. An 84dB speaker needs roughly four times the watts of a 90dB speaker for equivalent volume. This is why a 7-watt 300B amp sounds magnificent driving 97dB horn speakers and hopeless driving modern bookshelf speakers in a medium room at realistic levels.

It's also why I'd gently push back on the advice to always buy the most sensitive speaker you can find purely to enable low-powered valve amplification. Some of those high-sensitivity designs involve tradeoffs in bass extension, dynamic range, or midrange colouration that you might not want. Match the speaker to your listening habits first, then find the right amplifier topology for that speaker. Not the other way around. The best standmount speakers guide covers this tradeoff honestly across different sensitivity ranges.

Which topology for which situation

Valve amplification makes the most sense when: you have high-sensitivity speakers (90dB and above, ideally more); you listen primarily at moderate levels; you value the ritual and the tactile engagement of the equipment; and you're drawn to genres where harmonic texture matters — jazz, folk, classical, acoustic music. It also makes sense if you genuinely enjoy the maintenance cycle. Retube a valve amp properly and it sounds like a new piece of equipment. There's something satisfying about that.

Solid-state Class A or Class AB makes sense when: your speakers are below 89dB sensitivity, have complex or low impedance loads, or you listen at high levels in a larger room. It's also the sensible choice if you want predictable, repeatable performance without any maintenance considerations. Most modern speakers are designed with solid-state amplification in mind.

Class D makes sense when: efficiency matters (apartment listening, hot climate, rack-mounted system); you want serious power reserves without paying solid-state Class A prices; or you're building a streaming system where the digital integration in modern Class D integrated designs is genuinely convenient. Don't rule it out on principle. The best current implementations are excellent.

A note on hybrid designs

The hybrid approach — valve input or driver stage, solid-state output — is more than a marketing compromise. Done well, it captures some of the tonal character of valve circuitry while delivering the current capacity and low output impedance of solid-state. The McIntosh MA352 mentioned earlier is one example. The Pathos Classic One has been doing this for decades. They're not for purists on either side of the fence, but they're genuinely useful for people who want some of the warmth of valve presentation without the speaker-matching constraints.

I'll admit I got this wrong for years. I dismissed hybrids as neither-fish-nor-fowl for too long. Some of them are very good, and they make a real case for systems where speaker sensitivity is in the 87–90dB range — enough to benefit from some of the harmonic character of valve circuitry, not quite enough to drive well with low-powered pure-valve designs.

Whatever you end up with, spend time listening before you buy. Not in a showroom demo — at home, with your speakers, your room, your records. The topology that measures best on paper is not always the topology that makes you want to stay up until midnight playing one more side. And that, in the end, is the only measurement that has ever really mattered to me.

— Priya Anand, Vinyl & Valves Editor