Purifi amplifier modules

Purifi calls its amplifier platform Eigentakt, and Bruno Putzeys is quick to correct anyone who calls it revolutionary. In this Tech Talk, the Purifi co-founder explains why the Eigentakt amplifier is really the next step in a straight line running back to his earliest self-oscillating Class D designs — and why the breakthrough that made it possible was a piece of mathematics, not a new circuit topology.

Why the Eigentakt amplifier builds on Ncore rather than replacing it

Putzeys, whose amplifier design career stretches back to Philips, has favoured self-oscillating Class D amplifiers — designs where the feedback loop itself causes oscillation, rather than a separate oscillator driving a PWM modulator — since those early days, because they’re more robust and allow more loop gain from a similar circuit. What he didn’t fully understand until he started Eigentakt was why. The Ncore module was around six or seven years old by the time Purifi began drivers work, and Putzeys felt the timing was right to revisit the fundamentals rather than simply refine the existing circuit.

The mathematical breakthrough behind the Eigentakt amplifier

A Class D amplifier only ever responds to its input at the instant it switches, which makes it — despite being fully analogue — fundamentally a sampling system, similar in principle to a sigma-delta modulator in an analogue-to-digital converter. That analogy holds neatly for a uniformly sampled amplifier idling near silence. The problem starts as soon as the amplifier is driven hard in one direction: the switching moments stop being evenly spaced, alternating between long and short intervals in a pattern known as non-uniform sampling, for which no usable mathematical treatment existed. Working out how to analyse a self-oscillating loop under non-uniform sampling — and then use that model to optimise the loop’s components for stability all the way to clipping — is what actually separates Eigentakt from Ncore. The circuit differences that resulted are real, and patented, but they’re a consequence of the maths, not the starting point.

What actually causes the “veiled” sound of some Class D amplifiers

The payoff of that mathematical work is loop gain: around 80 dB across the entire audio band on Purifi’s amplifiers, more than 20 dB higher than was previously achievable while remaining stable. That extra gain suppresses distortion not just from the output stage but from the output choke and capacitors too — and the output choke turns out to matter more than most engineers assumed. Distortion created inside a ferrite inductor is small in magnitude but extraordinarily complex, and, crucially, it’s correlated with signal that passed through the amplifier moments earlier as well as the signal present right now. That defeats the usual psychoacoustic masking that hides ordinary, memoryless distortion — which is why a small number of listeners have long described a grit, hash or veil specifically over Class D amplifiers. Feedback, it turns out, is entirely capable of suppressing this kind of distortion too, once you understand it’s there.

How the Eigentakt amplifier actually sounds

Putzeys describes the result as warm and organic — an amplifier that doesn’t draw attention to itself, but sounds comparatively warm next to more conventional designs specifically because it removes so much of what was previously adding grain to the sound. Listening across a range of speakers, the recurring traits are tight, controlled bass, strong resolution, and open, organic high frequencies. He’s careful to frame this as a byproduct of removing distortion rather than deliberately voicing the amplifier: Class D was chosen purely for efficiency and compactness, and everything else in the design exists to tame what he calls “anything but amplifying music” until what’s left is a faithful, uncoloured signal path.