GaN-based Class D power amps

So nice to see a GaN thead with such a wide range of comments. I will add a few factual items here for the benefit of the readers and non circuit designers.
First of all Class D has been around since the late 1950's. The concept of Class D could not be fully realized until the semiconductor technology caught up with the Theory.

The theoretical assumption of Class D design is that the switching device – the transistor or tube, can switch instantaneously on and off. I will refer to the device as transistor from here on. If the transistor does not switch on and off instantly, a lot heat and current will be required. There is heat generated in the transistor as it goes from theoretical zero ohms to the off state of the transistor. This transistor acts a resistor which generates heat. The 2nd item is that the conduction will start between the positive voltage rail and the negative voltage rail as one transistor turns on and the other turns off. This causes more heat and more current to run, with potential damage to the transistors or very large amounts of heat and current flow. What Class D designs have done is to delay the turn on of 1 transistor until the other is completely off. At this point there is no processing while waiting for the transistor to settle down. This is called dead time. This is what folks call “Class D sound”. Which is quite true. One way to get rid of this was to use heavy feedback or feedforward to reduce this dead time effect. The side effect of heavy feedback (or feedforward) was sluggishness and loss of detail as the feedback loops tried to correct for this dead time.

Transistors inherently have a lot of capacitance. This is the cause of the slow turn off. The capacitance in the transistors (MOSFETs in particular) keep the charge going even though the voltage has been removed. This causes running at the top and over shoot. Hence the hard edge of Class D also. Class D typically operates at 400kHz and so RF design methodology has to now take place on the circuit boards, wires, components and the whole design. A simple item like having a trace on the top of the board and the bottom of the board created a capacitor that could now affect the sound.

The overshoot is a common cause of the hard edge found in a lot of solid-state equipment. That is because the speed of the solid-state transistor led to all sorts of parasitic capacitance's to come alive. As mentioned prior, board capacitance, component capacitance, even the via’s on a circuit board make a difference now. The overshoot would suppose the frequency be in the 100kHz or higher. RF design is a key factor in Class D and in making good solid stage equipment.

What GaN (Gallium Nitride on silicon or Gallium Nitride on Silicon Carbide provide is
1. Ultra-fast switching speeds
2. Almost no overshoot and ring
3. Minuscule parasitic capacitance on the device.
4. Much greater efficiency of operation.

I can only speak for the Merrill Audio amplifiers at this point.
1. There is zero dead time. So it is smooth and you cannot tell or scope any switching.
2. There is zero feedback. Since we don’t have to correct for distortion or overshoot or deadtime. This results in pure detail and air (assuming the track has all of this recorded in it). The difference is quite stunning.
3. There is miniscule parasitic capacitance and inductance. There are probably have more capacitance in interconnects.
4. The wasted energy is minimized and the design is very efficient. Hence the amplifier operates at a stable low temperature, keeping all components around it operating at a low temperature also. This temperature stability means the operating amplifier approaches theoretical design. There is not thermal compensation required or degradation of components due to heat or thermal gradients of components to worry about. While green is a side benefit, the actual benefit is stable operation of components. This is big as inductors, capacitors and resistors have heat coefficients (positive and negative) and have different values at different temperatures.
From a sonics standpoint, the new Merrill Audio ELEMENT series are easily an order of magnitude better than the VERITAS. But don’t take my word for it, go listen to them yourself. Take your amplifier and do a direct A/B comparison.

GaN transistors – first commercially available in 2010, will certainly become more pervasive as the costs come down. GaN material was used in LED’s for over 2 decades.

GaN transistors can be used for any class of amplifier. It is not limited to Class D or other switching designs. It is commonly used in the Cellular industry which was the first to take it up as an amplifier. This is because of the very fast switching speeds and power handling capability of the GaN transistor. It is excellent for operating in the Gigahertz range.

As the volume increase, production gets better, the price of the GaN transistors will come down and you should see them used ubiquitously, much like a silicon transistor today.

I don’t give out the details on my component designs, and that will hold true here also. And as noted in this thread, it matters what it sounds like, not whats in it. A Bugattis cost money and not everyone can afford one. So the best sound for what one can afford or is willing to pay is where all of us will land up.

Here are some of the early reviews and more to come from HiFi Plus, Enjoy the Music, Positive Feedback and so on.

Brutus Award , Positive-Feedback.com editor David Robinson, “After starting the listening sessions, I still didn't realize that they were Class D…at first. I really liked the initial impression that the Element 118 created in me: very fast transient response, clean, transparent, detailed, and lacking no dynamics with key recordings that I always use.
Then I walked over to the amps to assess the heat being generated, and was startled by the fact that the 118's were cool to the touch! "Damnation!" says I, "The bloody things must be Class D!"
They are, but with state-of-the-art design and execution that puts Class D into my top-o'-the-heap class of true reference amplifiers. Details will follow in my review, which I hope to have done in January or so of 2019. Stay tuned for that one…the Element 118 is serious stuff!”

You can also look up my thread on “Merrill Audio ELEMENT 118 on Tour” to see more reviews first hand. Some of the speakers used are high efficiency horn speakers. Also look for reviews by Kansas City Orchestra percussionist in other forums.

To summarize, it is not the topology that matters, it is the actual build, fit and design. The ultimate test is the listening test. Until one hears A/B comparisons and actually hears the component, it is very hard to speculate how good or bad something is by just the marketing material or chatter.

For the more detail inclined minds this is a good 20 min read.
https://www.qorvo.com/design-hub/ebooks/gan-for-dummies#form

Note having the fastest car does not mean you automatically win the race. It is but a tool of many tools required to win the race. GaN transistors is not a guarantee of good sound but one of a lot of items that will get you there. 

This is a side note to the Class D and cool running amps. 

One of the major benefits of running cool or cooler is that the components stay intact longer. A major benefactor of this cooler running amp it the Electrolytic Capacitor that is very common in power amps. The electrolytic capacitor offers higher density, higher voltage and higher capacitance for the size and price. A large number of capacitors and large capacitance is used for the power regulation, even if most of the regulation is done by semiconductors. 

The drying out would typically run about 20 % less after 10 years.  Heat accelerates the drying out of these capacitors reducing the performance faster.  One hopes electronics will last a few decades. Cooler running equipment have a better chance of this occurring. 

Of course, first and foremost are sonics. And it will have to start off very good to have this as a consideration.