There’s a disturbance in the force for amplifer design

August 30 2024, by Matt Murray, 11 min reading time

There’s a disturbance in the force for amplifer design

Usually, this column focuses on a ‘video’ aspect of A/V; however, for this issue of Connected, an intriguing audio-related subject has captivated my interest.

VOILÁ, AT MIDNIGHT, AVPRO WILL ALSO BE AN AUDIO COMPANY

On August 1^st, 2023, AVPro Edge's parent company, AVPro Global Holdings, acquired the home audio division of American manufacturer AudioControl (a mobile electronics division continues regular operation under its previous holding company).

While parsing the myriad complexities involved with an undertaking as daunting as AudioControl Pro (its unofficial rebadging), one of my most personally rewarding highlights included observing the collective sigh of relief and subsequent exhilaration expressed by the engineering staff upon learning the creative restraints they were previously tethered by would be removed and a green light for innovation given. The electronic circuit design team detailed how they came to embrace and continued to remain excited about the ongoing exploration of an amplifier topology centered on an emerging yet relatively enigmatic output stage semiconductor architecture based on a transistor technology called gallium nitride (chemical formula: GaN).    

CLASS ACTION

The AudioControl model lineup is delineated by designs from three amplifier classes: One unit is a standard Class-AB design, with all others Class-D or Class-H.

Of those configured in Class-D, one series features the first AudioControl amplifier range to utilize a GaN output stage.

GaN, the topic of this article, has issued its prospectus at being the first disruptor technology in the seventy-plus years of solid-state amplifier design and is well on its way to manifesting that destiny.

SILICON AGE(-ING)

Silicon has been the technological backbone of the semiconductor world since Morris Tanenbaum and colleagues at Bell Labs first developed a working silicon transistor on January 26, 1954. Silicon displaced early semiconductor materials such as germanium when an oxidation process, discovered accidentally at Bell Labs while coating a silicon wafer with silicon oxide to act as an insulating layer, went on to become a method critical to semiconductor development which led to the capability for mass production. During the superseding seventy years, silicon’s ground-breaking capabilities have seemingly been pushed to their theoretical limits. During this period, the Silicon Metal Oxide Silicon Field Effect Transistor (Si MOSFET) enabled the most significant advancement in solid-state audio amplifier design.

MOSFETS AND AUDIO AMPLIFIER DESIGN

I want to touch on amplifier classes where MOSFETs are utilized, first to show their versatility in output stage design and then to illustrate how inherent limitations have steered many amplifier designers to GaN technology.

When unconstrained by cost, many high-end solid-state audio amplifiers use MOSFETs in Class-A designs, where tubes once reigned supreme and, in the current renaissance the industry is enjoying, have regained prominence along with vinyl playback.

MOSFETs are commonly found in Class AB, G, and H, and Class D switching amps are almost exclusively MOSFET-based. In Class D amplifiers, ideal output power bridge circuit characteristics are low voltage drop with fast on and off switching times and low stray inductance. MOSFETs with fast switching times, compared to other semiconductor types such as insulated-gate bipolar transistors (IGBT) and bipolar junction transistors (BJT), have proved to be the best option for Class D amplifier designs.

CLASS WARFARE

Drawbacks to Si MOSFETs include using a fair amount of negative feedback when implemented into amplifier designs with switching characteristics.

Class-A designs are excluded as the positive and negative portions of the waveform (the audio signal) are effectively regulated by a single device (or series of devices in parallel), be they tubes or transistors. Known as single-ended or push/pull amplifiers, the waveform is conducted in its complete 360-degree positive and negative cycle, eliminating crossover distortion (the single device handles both the high “push” side of the waveform and the low “pull” side).

A 'bias' current is constantly applied, ensuring the entire waveform is reproduced linearly within the output device's range, with the device passing current 100 percent of the time. A by-product of these constant bias designs is continuous heat. A 50-watt constant bias Class-A amplifier with no signal still outputs 50 watts (while likely drawing 100 watts from AC mains power).  

Class AB is a switching design, dividing the waveform into halves for efficiency.    

While also a push/pull arrangement, the bandwidth of the waveform is extended to overlap at 180 degrees up to 200 degrees. Crossover distortion lowers to a point where it is inconsequential, but this elimination process requires using negative feedback. A design advantage to class AB, unlike class A, is that class AB varies bias, drawing fewer watts during the idle time of the waveform. A 50-watt class AB amplifier might keep the output devices active with only 10 watts of quiescent bias current, bolstering efficiency. When the voltage of the incoming signal rises, the amplifier ‘swings’ its output up to 50 watts.

Class G and Class H are technically variations of class AB using voltage rail switching and rail modulation, which improve efficiency through reduced power consumption. In conditions where the voltage load of the incoming signal lessens demand, the amplifier utilizes a lower rail voltage than a comparable Class-AB design. The amplifier switches to its secondary, high-voltage rail when the incoming signal dictates higher power demand.

More costly due to parts count and complexity, class H designs use high-current-capable MOSFETs to improve power transfer while minimizing heat levels. Some Class-G amplifiers may employ more output devices than Class-AB designs, with one pair of devices associated with the low-voltage rails and an additional pair reserved to boost power through the high-voltage rails.

Class-H mode expands upon Class-G, with the boosted voltage entirely variable, compared to a fixed threshold trigger point. This enables the amplifier to operate with sustained headroom when required while maintaining low distortion and maximum output efficiency.

D IS NOT FOR DIGITAL

While class-D amplifiers equipped with digital inputs exist, class-D is NOT a digital amplifier topology; rather, the ‘D’ represents the following ordered letter in enumerated amplifier types. 

Two distinctions may be used to define the digital aspects of class-D amplifiers. Class-D amplifiers use a digital technique called pulse-width modulation (PWM), where a small analog input voltage is subsequently amplified to the power levels required to drive loudspeakers. Pulse-width modulation is a misnomer that leads many to presume class-D means digital; however, ADC quantization never occurs. Voltage, current, and the element of time (pulses) are physical quantities of the analog domain, with Class D amplifiers also using analog-based global negative feedback and error correction.

Although a class-D amplifier holds a power consumption advantage over almost every other amplifier classification, it still has the potential for problematic fidelity, primarily due to switching distortion.

Though the original concept for Class-D audio amplification dates to the ‘50s.

It was hoped that audio signals could be amplified by output devices that did not operate in a linear gain mode but acted instead as electronic switches.

Class D performance has traditionally suffered from semiconductor switches lacking the optimized performance for open-loop linearity, relying historically on MOSFETs as its de facto device technology. Si MOSFET switching speeds have limited the audio performance designers hoped to have been able to achieve. 

Slower switching speeds contribute to higher distortion, requiring negative feedback as a class-D distortion offset. This is one of the reasons critical listeners have never widely accepted these designs, as Si MOSFET-based amps utilize a generous amount of negative feedback. Circuit topologies for most Class D designs remain Si MOSFET-based, though recent developments in semiconductor design are changing that paradigm.

WHAT IS GALLIUM NITRIDE?

Here’s a textbook-type description: Gallium nitride is a direct band gap semiconductor comprised of a binary compound substrate with a wurtzite hexagonal crystal structure. It is manufactured through a process known as metal-organic chemical vapor deposition (MOCVD), in which gallium and nitrogen are combined to form the crystal.

The following is a shortened explanation and somewhat of an oversimplification of semiconductor design and operational characteristics: Various solid materials are composited during semiconductor manufacturing, each with individual energy-band electron characteristics. Variations in energy-band structure are responsible for the disparities in material selection when designing semiconductors. Semiconductors have two energy bands with defined territorial edges: A Valence edge and a Conduction edge.

OK, take time for a deep cleansing breath…

MIND THE GAP

An electron-free isolated area separating the two edges is called the band gap. Electrons flow through the band gap, jumping from the valence band to the conduction band to create energy transfer. A semiconductor’s band gap rating, expressed as electron volts (eV), reflects the minimum energy necessary for electron transfer to occur. It directly relates to the substrate material's innate capability to conduct electricity and is the best distinguishing identifier between semiconductor classifications.

One of the most significant advantages of GaN over silicon is its bandgap, which gives it various electrical properties that equip it for higher-power applications with greater reliability. Gallium nitride has a 3.4 eV bandgap compared to 1.12 eV for silicon and MOSFETs, making it far more capable of supporting high-voltage circuit designs before failure than silicon. The wider band gap for gallium nitride means it can sustain higher voltages and higher temperatures than silicon MOSFETs, which is crucial for economizing amplifier size and thermal handling.

Gallium nitride is not a recent discovery, with early research dating back to the 1930s. The technology lay dormant until the synthesis process was reexamined and refined for use in high-brightness light-emitting diodes. Gallium nitride is responsible for the disc-reading blue light in Blu-ray machines.

Adapting the material to Field-Effect Transistor (FET) technology, a GaN metal-semiconductor field-effect transistor (MESFET) was shown in 1993. GaN entered the audio world and started to earn its disruptor status, particularly for High-End audio, in 2010, when the first GaN enhancement-mode transistors were designed to replace power MOSFETs in critical applications requiring high switching speeds and efficient power conversion.

GaN-ing an Advantage

GaN-based class-D amplifiers aim to achieve accurate, large-signal reproduction from a small signal source efficiently, distortion-free, with little energy loss. Among gallium nitride’s highlights are lightning-fast switching speeds and dramatically reduced ‘On resistance’ compared to even the fastest MOSFETs. Switching speed correlates to how steep the ‘on’ or ‘off’ slope of the switching waveform is. The steeper, the better, since that translates directly to lower distortion. GaN FET devices used in class-D amplifier designs can switch up to 100 times faster than silicon MOSFETs. Digital square waves produced by GaN-based FETs are closer to what is ideal and nearly perfect compared to what can be achieved by silicon, containing all the sound and harmonics from the incoming signal. Measurement of total harmonic distortion (THD) is an industry-standard metric. GaN FET power transistors in a class-D amplifier design consistently demonstrate a ten-fold reduction in THD compared to Si MOSFETs. This is due to the faster-switching capability and shorter ‘dead time’ (when no switching occurs) for GaN devices, lowering crossover distortion with higher signal-to-noise ratios.

When the waveform is filtered back into a sine wave through analog filtering at the output stage, the musical sine wave is identical to the incoming signal with no additional distortion. MOSFET-based Class D amplifiers require a large amount of negative feedback to reduce this distortion, affecting musicality and bass response. 

Speed is only one advantage of GaN transistors. Thanks to fewer passive effects from capacitance, resistance, and inductance, GaN transistors have inherently lower noise and can operate at much higher frequencies, providing fast, clean switching transitions. Switching frequencies capable of several MHz and the extremely low switching ‘dead time’ make GaN ideal for class D amplifiers.

Silicon switching times are possible down to 20 ns, with GaN measuring four ns. Semiconductor dead time creates distortion in class D designs. 

Gallium nitride, with better efficiency than silicon and better thermal stability, is perfect for class-D amplifier designs requiring sustained high-power handling under brutally demanding impedance loads while operating at higher voltages than Si devices. Gallium nitride is a future-proofed material for the semiconductor world and is poised to elicit a seismic shift in sound quality-centric audio amplifier design. As GaN becomes more prominently used, it undoubtedly will significantly improve the amplifier art; however, it cannot do that in and of itself. Every aspect of amplifier design must be carefully adhered to.     

• Share on:
• WhatsApp
• Deel
• X
• Pin it
• Messenger
• Email