This is the second part of a Tech Talk trilogy on Class D amplification with Bruno Putzeys — the engineer behind Hypex‘s UcD and Ncore amplifier modules, and co-founder of Purifi Audio. Having traced Class D’s decades-long prehistory in the origin of Class D amplifiers, this conversation moves to the current landscape of Class D amplifiers — which, as Putzeys tells it, is a considerably less dramatic story than the marketing around “digital amplification” usually suggests.
The current landscape of Class D amplifiers: self-oscillating designs became the standard
Since roughly the mid-2000s, the amplifier architecture that has quietly become the accepted standard across serious Class D design is the self-oscillating, global-feedback topology — the conceptual descendant of the circuit Putzeys built into UcD and later refined into Ncore. It’s the same basic approach used, in various forms, by manufacturers like Ice Power, Pascal Audio, Hypex, and Purifi. What used to be a niche, hard-to-get-right technique is now simply how a competent Class D amplifier is built, which is itself a sign of how thoroughly the “schoolbook” open-loop approach lost out once engineers understood why it had been holding the technology back.
Gallium nitride: useful, not revolutionary
Gallium nitride (GaN) transistors get discussed constantly in current Class D marketing, and Putzeys’ take on them is more measured than the hype suggests. GaN FETs are genuinely easier to drive well than traditional silicon MOSFETs, which is a real practical advantage for designers. But he’s clear that easier-to-drive is not the same as fundamentally superior, and switching a GaN stage faster than necessary doesn’t buy better sound — it just wastes energy as switching losses, with no corresponding benefit. In other words, GaN is a useful tool for making a good self-oscillating design somewhat easier to execute, not a separate technological leap on top of what global feedback already achieved.
His own lineage: UcD, Ncore, Eigentakt
Putzeys traces his own contribution to this landscape through three successive amplifier platforms: UcD, which introduced the global-loop self-oscillating approach; Ncore, which refined it further at Hypex; and now Eigentakt, developed at Purifi, representing the current state of that same underlying idea after two more decades of engineering refinement. Each generation kept the same fundamental architecture — comparator, output stage, low-pass filter, feedback loop, no separate clock — while pushing distortion and noise performance progressively lower.
A simple way to actually test an amplifier
Asked how he evaluates a Class D amplifier’s real-world performance, Putzeys describes a deliberately unglamorous methodology: record the amplifier’s input signal and its output signal, then compare the two by ear. It’s a technique that cuts straight through marketing claims and spec-sheet numbers — if the recorded output doesn’t sound identical to the input under normal listening conditions, something in the design needs fixing, regardless of what the datasheet says.
Amplification as a commodity — and the accidental “input buffer” trend
Putzeys is blunt that amplification itself has become a commodity: a well-executed Class D module is now cheap and widely available enough that it no longer functions as a meaningful point of differentiation on its own. One unintended consequence of that shift, which he traces back to his own work, is the now-common practice of pairing a commodity amplifier module with a dedicated input buffer stage — a trend he essentially started without initially setting out to create an industry-wide convention, and one that shows up across the very evaluation kits Purifi now sells.
The NAD M33 as a real-world example
As a concrete example of where this landscape lands in an actual consumer product, Putzeys points to the NAD M33 — an integrated amplifier built around this generation of self-oscillating, feedback-corrected Class D amplification, paired with the DSP and streaming functionality that modern buyers expect. It’s the kind of product that illustrates his broader point: the amplification stage itself is now a solved, commodity problem, and the interesting engineering happens in everything built around it.
Frequently asked questions
What defines the current landscape of Class D amplifiers?
Self-oscillating, global-feedback designs, in use since roughly the mid-2000s at manufacturers including Ice Power, Pascal Audio, Hypex, and Purifi — the conceptual descendant of the architecture Putzeys introduced with UcD.
Do gallium nitride (GaN) transistors make Class D amplifiers fundamentally better?
Not on their own, according to Putzeys. GaN FETs are easier to drive than traditional MOSFETs, but switching faster than necessary just wastes energy as switching losses without improving sound quality.
What is Putzeys’ own amplifier product lineage?
UcD, then Ncore at Hypex, then Eigentakt at Purifi — three successive generations of the same self-oscillating, global-feedback architecture, refined over roughly two decades.
How does Putzeys test whether a Class D amplifier sounds accurate?
By recording the amplifier’s input and output signals and comparing them by ear — a straightforward test that bypasses spec-sheet claims and checks whether the amplifier is genuinely transparent.
Why does an “input buffer” now commonly accompany commodity amplifier modules?
Putzeys traces this now-common pairing back to his own work — as amplification itself became a commodity, adding a dedicated input buffer became a way to preserve signal quality ahead of the power stage, and the practice spread industry-wide.
