GaN Amplifiers – A Genuine Technological Leap, or Just Another Fashionable Acronym?

When Class D No Longer Means What It Did Twenty Years Ago

The hi-fi world loves new acronyms. All it takes is a new technology and a well-sounding English term, and suddenly it seems as though every older component needs to be replaced.

GaN is in exactly that position right now.

Those three letters are appearing on more and more amplifier specification sheets, while manufacturers promise higher power, lower heat generation, faster operation and better sound quality. One of the latest examples is the HiFi Rose RA80 monoblock, which, according to the preliminary specifications, delivers 500 watts into 8 ohms, 1,000 watts into 4 ohms and peak output of up to 2,000 watts into 2 ohms. According to the manufacturer's current information, its release is expected in autumn 2026, although the final price has not yet been announced.

Those are impressive figures.

But I am not interested solely in how many watts an amplifier claims on paper. A much more important question is what is happening inside the chassis, how GaN differs from conventional silicon transistors, and whether any of this can actually be heard during music playback.

Because GaN is not magic. Used properly, however, it can represent a genuine step forward.

What Does GaN Actually Mean?

GaN stands for gallium nitride. It is a semiconductor material that is increasingly being used in power electronics as an alternative to conventional silicon.

Inside an amplifier, GaN FETs or GaN HEMTs are not usually found in the volume control, the input selector or some kind of special "sound-enhancing" circuit. They are primarily used in switching power stages, where the devices must turn on and off extremely quickly.

GaN devices can operate with lower gate charge and output charge, faster switching, and without the reverse-recovery behaviour associated with the body diodes of conventional silicon power MOSFETs. These characteristics can enable lower switching losses and higher operating frequencies.

In simple terms: the transistor can perform its task more quickly and more precisely.

That may not sound particularly romantic at first.

But in a Class D amplifier, this is exactly where a great deal can be gained.

Class D Does Not Mean Digital

This is worth clarifying immediately.

A Class D amplifier is not necessarily a digital amplifier. The letter "D" is not an abbreviation of the word digital; it is simply the next letter in the naming sequence of amplifier classes.

The output transistors in a Class D power amplifier do not operate continuously in a linear region, as they do in a conventional Class A or Class AB design. Instead, they switch on and off at very high speed. The musical information is carried by the duty cycle of the switching waveform, while an LC filter placed before the loudspeaker removes a significant proportion of the high-frequency switching components.

The resulting signal at the loudspeaker terminals is still analogue.

The main advantage of switching operation is efficiency. When the power transistor is fully on, the voltage drop across it is low. When it is off, almost no current flows through it. As a result, far less energy is converted into heat than in a linear output stage that spends much of its time partially conducting.

This is why a Class D amplifier can be smaller, lighter and cooler while still delivering substantial power.

Where Do the Problems Begin?

In theory, the switch is perfect.

At one moment it is fully on, and at the next it is fully off. In reality, however, a transistor requires time to transition between those two states.

If the upper and lower transistors in a half-bridge conduct at the same time, a direct current path can briefly form between the supply rails. This is known as shoot-through, and it must obviously be avoided.

For this reason, the control circuit leaves a very short pause between switching the two transistors. This is known as dead time.

Dead time is necessary, but it is not free. The longer it is, the more error it can introduce into the output signal. If it is too short, the risk of shoot-through increases. If it is too long, linearity can deteriorate and distortion may rise.

The switching speed, gate charge, internal capacitances and body-diode behaviour of silicon MOSFETs all determine how precisely this process can be controlled.

And this is where GaN comes into the picture.

What Does GaN Do Better?

One of the most important advantages of GaN is that switching transitions can be faster while less energy is lost during the transition.

This can provide benefits in several areas.

Shorter dead times may be used, switching losses may be reduced, the switching frequency may be increased, the components in the output filter may be made smaller, and more power may be fitted into a smaller space.

EPC specifically highlights lower losses, reduced heat generation and higher switching speeds in Class D audio applications. Infineon promotes GaN transistors for similar reasons in its own high-power Class D reference designs.

There is, however, something important to understand.

Faster switching does not mean that the music itself becomes "faster." A drum hit does not become more precise simply because the specification sheet lists a higher switching frequency.

A faster semiconductor merely gives the designer more freedom.

What the designer does with that freedom depends on the quality of the entire circuit.

A Higher Switching Frequency Is Not an End in Itself

The switching frequency is often discussed as though a larger number automatically means better sound.

But there are compromises here as well.

One of Infineon's 750-watt GaN reference amplifiers was designed to operate at a default switching frequency of 1 MHz. This allows smaller output inductors and a more compact layout. The same reference design can also be configured for operation at 400 kHz. In that case, larger inductors are required, but switching losses and the thermal load on the GaN devices may be reduced.

In other words, there is no single perfect switching frequency.

The designer must consider efficiency, distortion, temperature, the output filter, electromagnetic interference and stability at the same time.

This is where real amplifier design begins.

GaN Does Not Fix Anything Automatically

A poor amplifier can still be built around an excellent transistor.

GaN does not automatically solve a weak power supply, a badly designed printed circuit board, poor grounding, noisy control circuitry or an incorrectly dimensioned output filter.

In fact, faster switching edges can make a design even more sensitive to track length, parasitic inductance and component placement. A layout that still works with a slower silicon transistor may produce severe overshoot, ringing or electromagnetic interference when GaN devices are used.

This is why the following are especially important in a GaN amplifier:

• suitable gate-driver circuitry;
• carefully designed PCB layout;
• a stable, low-noise power supply;
• the feedback arrangement;
• the output LC filter;
• overcurrent, thermal and DC protection.

The GaN label on the front panel therefore tells us only what type of semiconductor has been used.

It does not tell us how well the amplifier as a whole has been designed.

What Can We Hear From It?

This is the most difficult question.

GaN amplifiers are often described using words such as fast, clean, transparent, dynamic, controlled or detailed.

Those may be genuine listening impressions, but they should not automatically be attributed to the semiconductor material itself.

There is no single "GaN sound."

In the same way, there is no single Class AB sound. A vintage Japanese integrated amplifier, a modern high-end power amplifier and a professional studio amplifier may all operate in Class AB, yet their sonic character can be completely different.

The technical advantages of GaN can make lower distortion, reduced noise, more precise switching and improved high-frequency performance possible. But how much of that reaches the loudspeaker is determined by the entire circuit.

That is why I would not ask simply whether an amplifier uses GaN.

I would ask: How have the power supply, feedback, output filtering, thermal management and interaction with the loudspeaker's changing impedance been handled?

Not All GaN Amplifiers Are the Same

There are several different approaches on the market.

AGD Productions has built an entire high-end amplifier range around GaN output stages. Its smaller Tempo model, for example, delivers 2 × 200 watts into 4 ohms, and the manufacturer has designed its power module as a replaceable, upgradeable unit.

Orchard Audio's Starkrimson amplifiers also use GaN output stages, typically operating at around 800 kHz.

The Peachtree Audio GaN 1 follows a different philosophy. It accepts a direct digital S/PDIF signal, contains no conventional separate DAC or analogue preamplifier, and relies on the connected digital source to control volume. The amplifier is rated at 200 watts per channel.

HiFi Rose takes a more visually striking and feature-rich approach with the RA180, RA280 and the new RA80. The manufacturer describes the RA180 as an advanced Class D amplifier based on its own GaN technology.

All of these are GaN amplifiers, but they do not represent the same system philosophy.

That is why it is unwise to draw far-reaching conclusions from those three letters alone.

It Also Matters Where the GaN Is Used

GaN may appear in the output stage that drives the loudspeaker, but it may also be used only in the switch-mode power supply.

Those are not the same thing.

A GaN-based power supply may be more compact, faster and more efficient, while the amplifier's output stage may still use conventional silicon transistors.

That is not necessarily a problem. The amplifier may still be excellent.

It is simply worth checking exactly what the GaN label means on the manufacturer's specification sheet.

Marketing materials sometimes like to blur the difference between technically distinct solutions.

We Still Need to Be Careful With Power Ratings

New technology does not change the fact that amplifier power figures should be compared under equivalent conditions.

It matters:

• at what load impedance the measurement was made;
• whether one channel or both channels were operating;
• at what level of distortion the figure was specified;
• whether it is continuous or peak power;
• how long the amplifier can maintain the stated output;
• what mains supply and cooling conditions were used.

Peak power looks impressive in a brochure, but it is not the same as continuously available output.

A short-term 2,000-watt peak figure does not tell us how the amplifier will behave with a difficult loudspeaker load for half an hour. The power supply and the cooling system still cannot be replaced by acronyms.

From a Service Technician's Point of View

From a repair perspective, GaN technology is both interesting and demanding.

Lower heat generation is good news in theory. Heat is one of electronics' greatest enemies: it accelerates capacitor ageing, stresses solder joints and affects the long-term life of virtually every component.

A cooler-running amplifier may therefore have a genuine long-term advantage.

On the other hand, these power stages are often built on densely populated, multilayer PCBs. GaN devices can be extremely small, and the driver circuit is closely linked to the geometry of the board itself. This is no longer necessarily a situation where we can simply find a transistor with similar specifications, fit it, set the bias current and send the amplifier on its way.

In a high-speed switching circuit, the PCB itself is part of the circuit.

A great deal therefore depends on whether the manufacturer provides schematics, replacement parts, complete amplifier modules or long-term service support.

AGD's replaceable power module is a good direction. Not because it can never fail, but because the design takes future repairability and upgradeability into account from the beginning.

A fully bonded, undocumented, proprietary module is far less reassuring.

What Would I Look at Before Buying?

I would not choose on the basis of the GaN label.

I would look at how the manufacturer specifies continuous output into 8 and 4 ohms, whether 2-ohm operation is supported, and whether meaningful measurement data is provided.

The design of the power supply, cooling system, protection circuitry, input sensitivity and level matching would all matter to me. I would also want to know how the frequency response changes with different loudspeaker loads.

I would check the warranty, the service network and whether a replacement output module is likely to be available in the future.

Finally, I would listen to the amplifier with my own loudspeakers.

Because no matter how attractive the measurements may be, an amplifier still has to work as part of an entire system.

Will GaN Replace Class A and Class AB?

I do not think the story is that simple.

Class A amplifiers will continue to have their own engineering philosophy and sonic appeal. Class AB is not going to disappear either, because it is a mature, well-understood technology that can still produce exceptionally good and durable equipment.

What GaN demonstrates is that Class D no longer has to be synonymous with cheap, lightweight and compromised amplification.

It can now be used to build serious high-end power amplifiers, high-output integrated amplifiers, active loudspeakers and professional multichannel systems.

Not because GaN automatically makes everything better.

But because it pushes several previous technical limitations much further away.

Octonano Opinion

I do not see GaN amplifiers as either a miracle or a passing fashion.

This is a genuine, technically meaningful semiconductor technology that is particularly well suited to high-speed switching amplifiers. It can help achieve lower losses, shorter dead times, higher switching frequencies and greater power density.

But that does not mean every GaN amplifier will be good.

Sound quality is still determined by the complete design: the power supply, control circuitry, feedback, filtering, PCB layout, mechanical construction and interaction with the loudspeaker.

For me, the most important question is therefore not GaN or silicon.

The real question is whether the engineer understood what they wanted to achieve with it.

A well-designed Class AB amplifier may remain better than a poorly designed GaN product for many years to come. A genuinely well-engineered GaN power amplifier, however, can demonstrate that high output does not necessarily have to come with enormous heatsinks, heavy power consumption and the heat output of a small boiler.

GaN is therefore not a new name for good sound.

It is another very serious tool for engineers who genuinely know how to design an amplifier.

And that, in my opinion, is exactly how it should be viewed.

Author: Norbert Somogyi
Illustration: Octonano-Elektronika