Editor’s note: I am pleased to bring you this insightful tech blog authored by Steve Colino, Vice President, Strategic Technical Sales, Efficient Power Conversion Corporation and Skip Taylor, CEO and Founder, Elegant Audio Solutions, Inc. Enjoy.
Class-D audio amplifiers have traditionally been looked down upon by audiophiles, and in most cases, understandably so. Switching transistors for Class-D amplifiers have never had the right combination of performance parameters to produce an amplifier with sufficient open-loop linearity to satisfy the most critical listeners. This restricted the classical analog modulator Class-D systems to lower-power, lower-quality sound systems.
To accomplish the required headline marketing THD+N performance targets, Class-D amplifiers have had to resort to using large amounts of feedback to compensate for their poor open-loop performance. By definition, large amounts of feedback introduce transient intermodulation distortion (TIM), which introduces a ‘harshness’ that hides the warm subtleties and color of the music that were intended for the listening experience.
The silicon MOSFET has been the switching transistor of choice for Class-D systems for the past 25 years; and although producing a more efficient amplifier, they have been plagued by distortion due to imperfect switching, high on-state resistance, and very high stored charge (see Figure 1). The recovery of the stored charge dissipates power and causes ringing, which leads to additional distortion in each switching cycle and cannot be eliminated through the typical output filter approaches.
The GaN FET has less overshoot and ringing than most MOSFETs to drive the speakers with smooth, clean analog output from the PWM and they are smaller.
The switching imperfections and slower switching transitions associated with the silicon MOSFETs limits Pulse Width Modulation (PWM) switching frequency. In addition, the lower PWM switching frequency constrains the upper end of the audio bandwidth by defining the implementation constraints on the output filter. Higher PWM switching frequencies allow for a higher audio bandwidth, and hence higher-frequency output filters. As a side benefit, this higher-frequency output filter allows for smaller output filter components (especially, the Inductors) without compromising the sonic performance.
Block diagram of a basic switching Class D amplifier (Diagram from EETimes article ).
In addition to offering higher audio bandwidths (which is increasingly important for the new high-definition audio requirements), the increased PWM switching rates also allow for more moderate output filter slopes, which offer more linear performance without introducing higher levels of residual switching noise.
What if there were a transistor technology with switching so precise that it could produce a near perfect power representation of the small audio signal produced by the PWM Modulator, thus reducing (or fully eliminating) the need for these large amounts of feedback? What if the technology were so disruptive that the bandwidth of the output filter could be doubled for high-definition audio without fear of increased EMI/EMC problems? What if the switching technology of a Class-D audio amplifier could boast an “On” resistance and switching losses that were so low that power dissipation became negligible? That transistor technology is gallium nitride (GaN), and is available today from Efficient Power Conversion Corporation (EPC).
EPC’s enhancement-mode GaN (eGaN) transistors switch up to ten times faster than silicon MOSFETs, with ‘zero’ stored charge. The increased switching speed of eGaN FETs allows amplifier designers to increase PWM switching frequencies, reduce dead-time, and drastically reduce feedback; producing a sound quality previously limited to large, complex, heavy Class-A amplifier systems.
Demonstrating the superior performance of eGaN technology with their eGaNAMP2016 amplifier, Elegant Audio Solutions of Austin, Texas, has produced an amplifier capable of a continuous power output of 200 Watts into an 8Ω, or 400 Watts into a 4Ω speaker load, with THD+N as low as 0.005% and very low feedback.
Elegant Audio Solutions' eGaNAMP2016 Class D amplifier with audiophile level audio performance.
Moreover, this is done without the need for a heat sink and eGaN-based amplifier can plug directly into the standard amplifier implementation of many existing systems.
This is not the first Class-D amplifier to take advantage of the enhanced switching capability of eGaN FETs. Panasonic reintroduced their high-end audiophile Technics brand , with eGaN technology last year, and this amplifier is capturing the attention of audiophiles everywhere. With eGaN technology, high-definition audio requirements of bandwidth increasing to 96 kHz are easily accommodated without fear of the classical Class-D amplifier barriers. Increased bandwidth allows the subtle intricacies of sound to come through – loud and clear.
In the speaker-driving section of the amplifier, Technics has employed a tiny high-speed GaN (Gallium Nitride) FET driver with super-low resistance. 
Since increased audio bandwidth has traditionally allowed for unintended distortions to come through to the listener, the challenge of producing an even cleaner audio signal from the new Class-D systems is ever-present. Historically, increasing the audio bandwidth, with the associated increase in PWM switching frequency, has introduced two significantly prohibitive factors; increased heat due to the inherent inefficiencies of the MOSFET switch, and increased switching waveform distortion (including a larger contribution from dead-band). Thus, the task is to send a higher quality PWM switching waveform to the output filter.
To accomplish this task, a much more perfect switching waveform is required. But what about the distortion and noise each cycle brings and the additional heat generated? Will the net effect be a larger, more complicated system to assure uncompromised sound quality? With silicon MOSFET technology, the answer is a prohibitive ‘yes.’ With eGaN technology, the increased sound quality can be obtained without the added complexity and without the increased power dissipation. For a sound-bar or similar application example, in a 50 W high-definition audio amplifier with a 96 kHz output filter limit, eGaN FETs running at 2 MHz from a +32 V supply rail will dissipate less than 2 watts over the full-bridge output stage. That is less than half-a-Watt per FET switch, which is well within their thermal capacity. In addition, the higher output filter frequency makes output filter design less expensive and the higher switching frequency makes EMI/EMC compliance much easier to accomplish.
As noted, since the switching edges of the eGaN FET are cleaner, and introduce less ringing, Electromagnetic interference (EMI) is easier to manage. The wafer level packaging of eGaN technology results in very low inductance. With a good layout, overshoot and ringing can be virtually eliminated . With the reduction in both the output filter costs and the virtual elimination of heat sink costs, the eGaN FET-based high-definition audio system will not only sound much better, but will be much smaller, and have a lower system-level cost than the classic MOSFET-based systems.
A high-definition eGaN FET-based system with higher PWM switching frequency, reduced feedback, and higher bandwidth produces the sound that has the warmth and sonic quality that audiophiles demand – a great audio experience that can go places where even the best linear Class-AB systems dare not go. Look to high-definition eGaN-based systems for your home theatre, your car, your boat, and your portable wireless speakers, along with your high fidelity home systems.
Just as high speed switching eGaN technology has disrupted a myriad of other industries, including telecommunications, television, and automotive to name a few, eGaN technology is poised to disrupt the high-end audio world. Class-AB audio’s historic lesser child, Class-D, has now come of age with eGaN technology.
 J. Honda and J. Adams, “How Class D audio amplifiers work”