Anybody else waiting for GaN and Purify to have a baby?

I thought these eval boards were not optimized for audio? At least the first generation one.
In any case, I doubt these boards possess the design magic and proprietary GaN's that AGD has.

@art_boston 

They are meant to be for audio, but also to allow a designer to evaluate the GaNFET capabilities. One of the ways they show that off is the board is supplied without heatsinks- it can run full power into 8 ohms without damage. This is also why the power supply that comes with that kit is current limited (so the amp is limited to 114 watts/channel into 4 ohms) so as to not overheat the output devices- which they would do if putting out 400 watts (the devices can handle that power if heatsinked). Overall the product ticks the boxes set out by GaN Systems but if you are looking for a high end audio class D amp you're better off looking elsewhere. Not that this board is bad by any means, just that other products not meant for evaluation are better for audio.

Alberto PURPOSEFULLY used 400kHz in the first version because he noticed no improvement in sq beyond that. His amps were considered awesome with that frequency. The newer tubes now use 800kHz, purportedly offering increased resolution across the frequency board.By the way, the use of these frequencies certainly disproves George’s position of go 1.5MHz or don’t go at all.

As you increase the switching frequency you run into an issue with deadtime. I'm not going to go into what's actually going on here but I'll just say that the output filter plays an enormous role in the switching. So you may find that if you increase the switching frequency you may have more need for deadtime, not less! If you increase the filter frequency you get on a different part of its discharge curve and this will have a direct effect on the deadtime required. But the overarching issue is controlling noise and ringing as you increase the switching frequency!


@georgehifi  You might consider that because you heard one particular tube amplifier, that does not mean you've heard them all! I'm not a particular fan of SETs either, although I'm very aware of some of their strengths. I think their weaknesses outweigh their strengths and like you, I want more detail and impact than they seem capable.

The "distortion signature" is just one of an infinite amount of things that effect the sound.

It is true that phase shift can cause colorations. But beyond that almost anything else that results in an audible difference is likely due to how it affects the distortion.

"Designed properly".......now that is funny. There is no consensus in high end audio as to what that means. 

This is true. For me, 'designed properly' (which I hope was easily inferred from my prior post) simply means that the amp will have a distortion signature that allows it to be neutral (while otherwise allowing the amp to be usable and reliable....). I don't regard most solid state amps as neutral because their distortion signature includes unmasked higher ordered harmonic distortion which is audible as brightness and harshness; literally the reason that vacuum tubes are still around. They aren't 'designed properly' IMO.

If all you need is a certain "distortion signature"......then why make a new amp? There are already amps with that signature.....nothing new there.

This is true as well. However the reason to make a new amp is simple: the amps that have had the right distortion signature in the past are all tube amplifiers. None are solid state. Class D offers the ability to have a similar distortion signature, but at a much lower level, easily over a magnitude lower. This allows the amp to also be more neutral. A class D amplifier also does not need regular service, makes a lot less heat and is more compact. If you can offer all those advantages over tubes with the same smoothness and detail thru the mids and highs why on earth would you **not** do that?


The trick here is understanding how the measurements correlate with what we hear. This is rarely done in the spec sheets which give you very little information about how the circuit actually measures. This is why I'm talking about the distortion signature since it is something that can be easily measured with today's equipment and its predictable that if you get it right the amp will sound like music to the human ear.


Here are some aspects of a proper distortion signature:1) the THD measures the same at 100Hz, 1KHz and 10KHz (usually only the 100Hz distortion is shown in reviews and on spec sheets)2) the primary distortion components should be the 2nd and 3rd harmonics (which are nearly inaudible to the ear) in sufficient amount such that they mask the higher orders. A tube amp makes more higher ordered harmonics than almost any solid state amp, yet sounds smooth because the lower harmonics are masking the presence of the higher harmonics.


If you don't get these two things right the amp will not sound musical and organic.


The chief engineer at Scott once said "If it measures good and sounds bad, -- it is bad. If it sounds good and measures bad, -- you've measured the wrong thing."
It was true when he said it and its still true today.

That depends on the distortion signature, which seems to be more important than **how much** distortion is present. If "designed properly" (IOW having a similar distortion signature) I would expect the same sound quality too. This is simply because most of what we audiophiles describe as differences in amplifiers comes down to their distortion signature.

I don't think that is a good guess. Ease of heatsinking? Have you seen the tiny heatsinks that people use on the GaN's.

Look how tiny the heatsinks are on this new Orchard 250 watt into 8 ohm module are:

https://orchardaudio.com/starkrimson-ultra-module

picture is at bottom of opening scene.....just scroll through them....you will see.

The GaNs are much faster than the mosfet Bruno uses. Here is what Bruno and Lars of Purifi say:

https://6moons.com/audioreview_articles/merrill/2/

Yes, I saw that. And I read that bit, but its really not in conflict with what I mentioned and I agree with it.  At 500KHz the extra speed of the GaNFET just isn't a thing- sure you can run a tiny bit less deadtime. But Bruno is running a **ton** of loop feedback (the Purifi module's patents are all about how he does that) so the distortion he's getting really isn't affected by the small difference in deadtime he might get with GaNFETs. Its not cost either since that's not a thing anymore, or wasn't a year ago anyway. That is why I think its something else- and the body diode would be a good example of that. If I were him I don't think I'd be trying to educate anyone on what that's all about in an interview :)

If I had to guess I would say that Bruno prefers the ease of heatsinking his output devices, which is why he does not use GaNFETs. Back about 5 years ago there was a distinct improvement in switching speed offered by GaNFETs but that's all but gone with recent MOSFETs. So these days the design consideration might have more to do with the body diode of the device and how you heatsink it.