GaN-based Class D power amps

Regarding power output of the in production GaN-based class D amps, I’ve been criticized for posting what is from the manufacturers’ own websites. So this is my last attempt to compare their respective outputs:

ADG Vivace: 200W into 4 ohms with 0.1% THD using a 1kHz signal

Technics SE-R1: 300WPC into 4 ohms with 0.5% THD using a 1kHz signal

Merrill 118: 800W into 4 ohms (no THD or frequency signal data specified).

Every manufacturer, has a different approach about how to provide (a bigger) number, when related to the output power.
It is the main number that attracts first attention, and for the most part actually it means not very much, but yet that is why the temptation (thanks to the lack of industry standards) to find a point in the THD vs Frequency chart as favorable as possible, and get it as big as possible.  
We have chosen to provide a number, within the characteristics of the design of the GaNTube, that still has some more meaning, in relation to a power level that can be delivered within reasonable quality (low distortion) and generate a "ears-drum safe" sonic pressure, if ever used.
With a 100W input signal, a speaker with even low efficiency, let's say of 87db at 1W/m., can generate a pressure of almost 110db...so...

By the way, comparing the power specs, it seems to me that the only way to get to that power level, considering the GaN MOSFETs available in the market, both Technics and Merrill amps are likely adopting an internal bridge configuration, for the power stage.
In other words, with a bridge configuration, the power would have be 4x. (The load sees twice the voltage swing, and since P=V^2/R if V is double then P is 4x).
The Power stage integrated into the GaNTube is not in bridge mode, it is a single stage, and for whoever is curious, uses a custom integrated half-bridge GaN Mosfet made by IR (now Infineon).

Also, I would like to share something more. When the idea of the Vivace was conceived, so to speak, we choose to go with an unique design, both for the technical and the aesthetic one.
That decision was made because we wanted to convey the message that a ClassD amp can sound better than the best classic Tube amp, and actually look very nice too and different (not just like another metal box of exaggerated dimension).
The aim was in principle to liberate the Class D topology from the "stigma" it has been carry-over for way too long, and for sure now, at least with GaN, without any foundation or (de-) merit; actually all the contrary. And it seems it is finally happening.
In any case, what I found peculiar too, is that from all these comments, now so many, it seems only very few people actually have had the opportunity to do some true comparative listening tests.
By reading the comments, whoever has done it, more and more it sounds to me like someone that "...has seen the light" (sorry for paraphrasing St.Paul...but...)
So my suggestion is...type less and listen more...go get to listen to these amps!

AGD

 Fluoride in our (US) toothpaste is nasty, too. Feds have no problem feeding the health benefit bs line to us. Lol

Yes, but it is peddled as a benefit, whereas power consumption cannot be disputed

:((((

I should never have checked out this thread.

Reading Merrills' amp reviews/comments I am so disheartened that the commenters own amps (including Merrill's pre Element series) significantly more expensive than anything I can dream about being so much better. Even the AGD Vivacies (which also have outstanding reviews) are way out of reach 

@tweak1, Early adopters pay the heavy price. 

I’d like to see the raw output of a GaN switch in a class d amp configuration, and find out exactly how much of that hard slewing --consists of odd ordered harmonics.

As, if the device is incapable of being linear, then ..it it very likely going to produce large amounts of odd ordered harmonics.. theory seems to say that but the design of GaN may make this problem go away. This problem that exists in all transistors up to this time.

the more non linear the gain curve of a transistor, the more odd ordered harmonics it generates. The more linear the gain curve of the transistor, the more even ordered the harmonics in it’s distortions.

This is why Nelson Pass went out of his way to have set of SIT (V-Fet) transistors made by semi-south (a manufacturer since gobbled up by someone else).

V-Fets have linear gain curves, very much so, similar to a triode tube. This is a big deal, a very big deal and it becomes that, when a person hears it..for a long enough time to understand what is going on. like intelligence, it takes different amounts of time for a person to work through--to the hearing of it. Some might never get there (cable haters, etc).

This not speculative, it’s part of the human spectrum. Until one hears a v-fet signal reproduction system, they have literally..no idea. It’s all gobbledygook and a bunch of words, until then.(of course, they need to possess the wherewithal in their personal and unique hearing and brain... and the time spent listening--- to ’get it’)


The harder (or faster) a transistor turns on or off, in time, ie, hard slewing, the greater the set of distortions produced in level as compared to the pure on/off aspect.

Using that signal (from the GaN transistor) as the basis of the sine reproduction, means that the sine, or audio waveform recovered is going to consist of a notable amount of odd ordered harmonics, which is very unfavorable to human hearing.

The fact that the GaN transistor turns off and on.. fast... and does not generate heat ’so much’ ( in comparison to fets), means it generates lower harmonics, and that may be key here. This is the whole enchilada of my question set and points/data mentioned. Are those distortions odd ordered or even, what is the mix if any... and what are their levels compared to the main switch period itself?

We do indeed hear odd ordered harmonics better, but that would be the wrong direction. We hear them and we don't like them. They are uncomfortable and don't represent or mix with our reality. They stick out. Thus, false obscuring detail. The early transistor sound.

This is an important question, it is deeply connected to the class d amplifier issue, as the prior incarnations of class d all suffer from this problem. It is why some of us are not so fond of class d amplifiers. The signal is too close to being based on large amounts of odd ordered harmonic switching reflections. We hear it and it sounds like..well.. uncomfortable, dirty, screeching, obscuring, false details and generally..... unwanted.

If a GaN transistor will not trace a linear gain curve and will not do ’gain’ and only do switching, then it may be entirely odd ordered harmonics that are generated in this fast switching time?

Which is entirely wrong for audio.

This is a very important question set about the fundamentals of the move into GaN for audio.

Lets try not to do the wrong thing again as it is economically cool (new toy, wonderment, etc) to do so.

This is the sort of stuff we missed with digital that took us +20 years to get a grip on after the fact ...and really begin to fix the the damned mess. Twenty years of crap had to come first. Let’s not do this again. We’ve already started down that road with class D as it stands....