As an audiophile on a journey of perpetually better sound, I was left slightly disconcerted by class-D trailblazer Bruno Putzeys’s assertion, in my previous class-D article, that, as he put it, well-done class-D amplification was now so good that “the amplifier part is basically a solved problem”.

I understood the gist of what Bruno meant in the context of my article—that he believed class-D had become so good that it made more sense, in the pursuit of greater gains in sound reproduction, to focus on other aspects of the playback chain, such as system integration. It’s hard to argue with the man. The Ncore class-D modules and Kii speakers Bruno’s been instrumental in designing and making work together are accomplished examples of his integration philosophy. The thing with Bruno’s above comment, however, was that it could also be interpreted as meaning that, in his view, class-D technology had reached a certain plateau in performance where it couldn’t get much better.

So, I sent off an email to Bruno asking if that’s what he meant; was class-D about as good as it will ever get?

“Before going on, I’m assuming here that what we call a ‘good’ amp is one that minimally changes the sound,” he wrote back. “That’s important because quite a lot of amplifiers that are considered respectable fail quite miserably by that definition. One very good class-A amplifier, in my book, is the Boulder 2150. A Gryphon Apex or Pass Labs X amp, not so much. Their character may be pleasant to some listeners, and by that definition merit the term ‘audiophile’, but high fidelity is something else.

“So, whatever I have to say about class-D must be seen in that light,” he said. “And by that reckoning, class-D has now fully drawn level with the state-of-the-art. That doesn’t mean there’s no [room] for refinement. I [meant that] class-D is a solved problem because there are no fundamental hurdles left. No new technology needs to be developed for whichever refinements are still to come. There are only so many ways to skin a cat, and—to continue the unpleasant analogy—if the cat still has a few miniscule patches of fur left, that doesn’t mean that completely new skinning technology is in order. Class-D amplifiers will continue improving gradually, not because they have to catch up with class-A (that time is past), but because of the competition [between] class-D designers.

“So, expect refinements, not revolutions,” he continued. “In the case of standalone amplifiers, some of those refinements are simply in the bit that sits between the input connector and the class-D module. When Hypex first started selling class-D modules, I decided to allow our OEM customers to add their own input stage and claim they’d done something special with it. For the same reason, Purifi’s eval kits come with intentionally rudimentary input buffers. I somewhat underestimated the impact of that decision, as it’s now become a big chunk of class-D lore in audiophile circles, that the module is ‘just the module’ and the real magic is supposedly in the input stage. Still, I guess it was a good move, because this way even the kind of designer who wants to make a heavily coloured amplifier can use my modules and simply dial in their signature sound in the input buffer. And when people talk about the ‘implementation’ of a class-D module in an amp, that’s specifically what they’re referring to.”

He added: “Another reason I was moved to call class-D a solved problem is because current top-end class-D modules are now so good that in any sound system, irrespective of cost, any other link in the chain is much weaker. Even if future modules get better, the improvements are minuscule when compared to the impact of, say, the speaker. That’s also true of great class-A amps like the Boulder 2150. But why bother if a class-D amplifier that’s a fraction of the price and size will do just as fine? State-of-the-art class-D modules have [made] amplification good enough [to turn] ultra high-end audio into a commodity. Anyone looking for a problem to solve that really moves the needle in audio will find loudspeaker design a much more satisfying line of inquiry.”

I couldn’t imagine how high-end class-D amplifier manufacturers would reply to such an outlook on class-D’s future growth prospects, so I asked them.

“There are a few areas where class-D amplifiers are improving and could stand further improvement,” said Elegant Audio Solutions’ Skip Taylor, who responded on behalf of Peachtree Audio, which uses Skip’s GaN FET class-D modules in its amplifiers.

“Class-D amplifiers typically fall into one of two distinct topologies: analogue input and digital input,” he said. “Almost all the Class-D amplifier designs and amplifier modules, as well as amplifier integrated circuits (ICs), fall into the first category. Even amplifier ICs that claim to be digital input are analogue input, preceded by a DAC of some sort.

“Analogue input is the ‘legacy, historical’ method of designing a class-D amplifier. It allows for the implementation of classic negative feedback technology to improve the ‘measured performance’ required by today’s audio consumer. Most class-D amplifier modules available today use this analogue input topology. Unfortunately, the analogue input approach is becoming less and less [compatible with] today’s ever-changing digital source needs.

“Digital input, if done correctly, is a far more complex implementation,” he said. “And most of the attempts at providing a digital input class-D amplifier have fallen short of expectations, both in complexity and scalability. This is an area where additional improvement is needed and desired, as this method is far more applicable to today’s market.

“In addition to class-D amplifier topology are the output power stage devices,” he continued. “A huge percentage of today’s class-D amplifier designs, including modules, ICs and custom discrete designs, utilize the silicon FET [transistor] as a power device (an on-off current switch—ed.). While the silicon devices are getting better with each revision, they still suffer many of the same issues that are characteristic of silicon FETs. The newer GaN FET technology eliminates many of these characteristic anomalies and offers a much better solution for both audio and thermal performance. The GaN FET technology is being accepted, but slowly, mainly because silicon-based class-D amplifiers already have [such a foothold] in the marketplace and there are limited resources to make the change. Power device technology, which allows for a more optimal switching output signal, is another area where improvement is still possible.

“These [improvements],” he said, “are facing obstacles because of the ‘installed base’ of current class-D modules and the fact there are ‘work-arounds’ for their inefficiencies and anomalies. These things fall short of making class-D ‘the best that it can be’.”

Orchard Audio’s Leo Ayzenshtat had this to say:

“Do I think class-D amplification is about as good as it will get? The short answer is no but let me elaborate. In some respects, class-D amplifiers have already exceeded the capabilities of human hearing. For example, noise and distortion are already below the noise floor of the quietest listening rooms. But there are other factors that determine sound quality, like phase shift, transient response, intermodulation, and others. Wide bandgap semiconductors like gallium nitride (GaN) ones, which have been used in all of Orchard Audio’s fully analogue Starkrimson amplifiers since 2017, have allowed some of these qualities to significantly improve class-D amplifier performance. One of the most important class-D design aspects is the output filter. GaN technology has made it possible to reduce the [detrimental] impacts of this filter.

“Implementation is the most important aspect,” he continued. “It’s extremely important not only for the amplifier module but for everything that goes around it. The power supply [has a major] impact on sound quality. It needs to be able to provide the amplifier with the current it needs, and if it can’t do so, the amplifier’s overall performance will suffer. While I strongly believe that our amplifiers have one of the best class-D implementations on the market, there is room for improvement.

“We are approaching a point when the amplifier will exceed the capabilities of human hearing, perhaps in the next 5 to 10 years,” Leo said. “The technology is making amplifiers more efficient and thereby smaller. You can expect amplifiers to shrink in size and weight. There is no longer a need for 100lb or even 40lb behemoths to achieve great sound—Orchard Audio’s 500Wpc stereo amplifier weighs less than 20lbs, and our class-D modules in the future will feature higher switching speeds, allowing for further reduction in phase shift, improved bandwidth, and reduced size.”

John Stronczer, Bel Canto’s founder and Chief Engineer, began his reply to me with a revelation: “When I first heard a Tripath demo board in 1999,” he wrote. “I realized that class-D was the future of audio amplification. It had the ability to communicate the musical message, especially through the midrange, in a way that traditional solid state class-A, A/B, and push-pull tube amps could not, but which our single-ended triode amps and these interesting little Tripath amps could. The efficiency and simplicity of the class-D output stage was not just a technical advantage but critical to the performance that could be achieved.

“I decided then that Bel Canto would pursue class-D amplifiers and leave behind the 20^th century technologies,” he said. “Developments from ICEPower, Bruno Putzeys’s Hypex, and Purifi, have pushed the class-D amplifier module well past the early Tripath days with improvements in power, distortion through the audio band, and especially noise, with dynamic range approaching 130dB. I now view the latest Ncore and Purifi modules as the modern-day version of the 845 triode tube. These latest amplifier modules are largely a blank slate that the final amplifier designer can use to create very compelling, high-performance audio products.

“While the best modern amplifier modules have low distortion and noise, and can deliver nearly unlimited power levels,” he said, “there remain many, many details in the product design that determine the final performance quality. System architecture, passive components, power supply design and implementation, voltage gain stages, source stages, and other design considerations all contribute to the end result. While the measurements are a critical indication of the potential of a given amplifier module, the ear still plays an all-important role in determining critical design and component details in the final product. Sometimes, seemingly tiny decisions will have an outsized influence on sonic performance and not all implementations using a given module will sound the same.

“While these latest amplifier modules do achieve extremely high levels of performance,” he added. “I am certain that development will continue and there will be new modules in the future. Technology is never static or else it dies.”