
From output-inductor behavior to stability at high levels, Purifi's technical work emphasizes modeling, feedback, and measurement of specific sources of error.

Danish company Purifi is widely associated with Eigentakt amplifier modules, but it also develops loudspeaker drivers. Its published work follows a common sequence: understand what limits performance, model the problem, verify a solution, and develop it into a product. Purifi
For an amplifier, a central question is how to maintain effective error suppression while the output stage is switching and the musical signal is continually changing in level.
The people behind the work
Purifi's team page identifies three co-founders: Bruno Putzeys, Lars Risbo, and Peter Lyngdorf. Bruno and Lars are both listed as CTOs; Peter is listed as an investor. Bruno's experience includes amplifiers and converters. Lars founded Toccata Technology, subsequently acquired by Texas Instruments, and his research interests include signal processing and mixed-signal systems.
Carsten Tinggaard brings a substantial loudspeaker background. The company lists him as co-owner and COO. He founded PointSource Acoustics and previously worked as Product Manager at Danish Sound Technology and CTO of Tymphany Denmark, with responsibilities covering the Peerless, Vifa, and Scan-Speak brands. These experiences connect amplifier, converter, transducer, and product-development work within the team. Purifi team biographies
Start with the output inductor
High-power Class D amplifiers commonly use output filters. Magnetic-core inductors offer practical advantages, but their response is not perfectly linear. Purifi's discussion of hysteresis highlights that a core's present behavior depends partly on its previous excitation, and the associated distortion does not necessarily fall rapidly at small signal levels. This Thing We Have About Hysteresis Distortion
Hysteresis can be understood as a response with memory: the same present excitation need not produce the same magnetic state after different preceding conditions. It cannot be described adequately by a single static inductance value.
This leads to a control question as well as a component question. If the filter introduces an error, can the control system correct it? Removing every magnetic core would not remove the engineering trade-offs; size, copper loss, stray fields, and current capability would still matter.
The objective is to reduce errors in the final audio output while choosing components and control behavior that work together.
Full output feedback and loop optimization
Purifi describes Eigentakt as bringing the power path, including the output filter, into the feedback loop. Its technical explanation also shows why increasing gain alone is insufficient. The behavior of a self-oscillating loop at large signal levels and near clipping must be predictable, requiring suitable models and numerical optimization. Combating Hysteresis Distortion
Numerical optimization describes a development method. It does not mean the music must be processed as digital audio inside the module during use. A circuit can be designed with computational tools while accepting an analog audio input.
Behavior near the limits
Good performance at low levels does not by itself establish good behavior near maximum output. As level rises, voltage swing, device current, and modulation range approach their limits. The feedback system has to operate through those changing conditions until the available hardware becomes the constraint.
Insufficient voltage headroom produces clipping. Excessive load current may activate current limiting. Neither boundary disappears because a small-signal distortion figure is low. Protection, the approach to overload, and recovery after the signal returns to range all remain relevant.
Evaluating a platform therefore means considering low-level performance, higher output, different loads, and behavior around clipping together. A user need not perform all these tests personally, but can check whether the technical information addresses them.
Three numbers that describe different things
Purifi's general Eigentakt page describes more than 75 dB of loop gain across the audio band. The second-generation 1ET6525SA datasheet specifies more than 85 dB and a corresponding 356 MHz gain-bandwidth product. The module's audio voltage gain is 12.3 dB. Eigentakt overview · 1ET6525SA datasheet, Electrical Specifications and Introduction
Audio voltage gain describes how much the input signal is amplified. Loop gain relates to the control system's ability to suppress error. Gain-bandwidth product is not the power devices' switching frequency. Treating these numbers as interchangeable would confuse input sensitivity, feedback, and switching behavior.
A relatively low module voltage gain also makes front-end design significant. A power stage capable of a large output voltage does not necessarily reach that output from any ordinary source. Buffering and additional gain must provide the required interface without introducing excessive noise or distortion.
Module families: distinguish architecture and application
The 1ET6525SA is a single-channel, analog-input, second-generation Eigentakt module. The 1ET9040BA also belongs to the second generation but uses a bridged output stage. Purifi quotes 750 W into 4 ohms at 0.1% THD for the 1ET9040BA. These are module capabilities under the manufacturer's conditions, not a statement of continuous output from a Maroon finished amplifier. 1ET6525SA · 1ET9040BA
The appropriate choice also depends on speaker impedance, required sound level, channel count, and cooling. More power and current headroom mainly help with higher output demands and more difficult loads. If a system already has adequate headroom, other requirements may matter more than a larger model number.
The input stage still matters
A low-error power stage does not make the rest of the amplifier irrelevant. If a buffer overloads first or input wiring introduces noise, the power module cannot automatically reconstruct the original music. Feedback can only correct errors within the relevant control path; interference entering as part of the input may be amplified along with the wanted signal.
In his discussion of low-gain buffers, Bruno addresses input impedance, the ability to drive feedback networks, and power-related interference. These requirements explain why integrated or discrete implementation should be chosen to meet a circuit's needs. Discrete Op Amps: When and Why It Actually Matters
Purifi offers modules and design-licensing arrangements, so products using its technology need not all have the same internal appearance. Technology identity, implementation, and complete-product specifications are separate questions. This series concerns Maroon's stated module-based approach. Eigentakt modules and licensing
From research to a repeatable product
Purifi's account of its development also covers facilities, production, laboratories, and the work of bringing technology into customer products. Manufacturing, quality control, and application support determine whether a circuit can serve customers over time. Purifi's company-development account
For a finished-amplifier manufacturer, Eigentakt provides a low-error power-amplification foundation. Input design, power supplies, thermal engineering, and complete-product verification remain necessary. Describing those responsibilities clearly is more useful than treating a technology name as a guarantee of every aspect of performance.
More in this series
How Much Amplifier Power Do You Really Need?
How Modern Class D Amplifiers Have Improved
Class D Does Not Mean Digital: Analog Inputs and Preamp Matching
Understanding Hypex: From UcD to NCORE and NCOREx
Bruno Putzeys: Bringing Class D to the Heart of High-End Audio
Browse the System & Craft Journal