Atma-Sphere Class D GaNFET Amplifiers

[SIY]

There have been numerous studies attempting to correlate distortion spectra with subjective preferences and there has been data presented that shows for some groups 2nd and 3rd harmonic distortion is agreeable to some listeners. This is often mentioned as a reason for the popularity of tube amps with this sort of harmonic distortion spectrum. No coincidence then if a tube amp manufacturer starts selling a class d amp with a similar distortion spectrum in order to please his tube lover customers.

Yet, at the levels we're talking about here (or in, say, Pass's amps), the manufacturers have not shown that their amps can even be distinguished, much less preferred (other than frequency response errors from the source impedance). The distortion spectrum thing is a nice sales point, but there's no actual evidence behind it.

 

[goryu]

Yet, at the levels we're talking about here (or in, say, Pass's amps), the manufacturers have not shown that their amps can even be distinguished, much less preferred (other than frequency response errors from the source impedance). The distortion spectrum thing is a nice sales point, but there's no actual evidence behind it.

Yet, at the levels we're talking about here (or in, say, Pass's amps), the manufacturers have not shown that their amps can even be distinguished, much less preferred (other than frequency response errors from the source impedance). The distortion spectrum thing is a nice sales point, but there's no actual evidence behind it.

My point is that there have been studies which show certain groups prefer certain distortion spectra. I agree that it is up to the manufacturers to prove that their products produce said spectra and that it is discernible.

 

[atmasphere]

Nor does he dwell in the universe of presenting evidence for these claims.

At least the engineering is excellent.

Thanks, regarding the former, the 'evidence' is all around us. I would start with the Radiotron Designer's Handbook (I looked this one up when John Curl mentioned it for the same reason I do) 3rd edition, published in the 1930s. In it, fairly early on you see that there was an understanding at that time that the ear was considerably more sensitive to the higher ordered harmonics than the lower orders.

That simple fact is what has kept tubes in business all these years. That tubes are still around is the second bit of evidence.

What is less understood is that the 2nd and 3rd (both of which are treated by the ear the same way, which is to say they are innocuous) can mask the higher ordered harmonics, similar to how this happens in a loudspeaker. So a tube amp, which generates considerably more higher ordered harmonics than most any solid state amp, won't sound bright and harsh on that account because the harmonics are masked. If its not clear at this point it should be: the ear/brain system assigns a tonality to all forms of distortion and it is most sensitive to the higher orders since it uses them to sense sound pressure.

No coincidence then if a tube amp manufacturer starts selling a class d amp with a similar distortion spectrum in order to please his tube lover customers.

Actually the amp was built because I came to the opinion about 5 years ago that if a manufacturer of amps in the high end audio market didn't figure out class D pretty soon, they were going to get left behind; this is gradually playing out in the high end audio market. We might be a little weird compared to a number of high end audio companies in that we actually have engineering capacity in house, using someone else's modules seemed to send the wrong message. We were issued a patent about a year and a half ago.

Personally, I prefer an amp that produces distortion below audibility. Producing "agreeable" distortion to appeal to certain listeners or to mask other less agreeable distortion doesn't seem to fit with the purpose of an amplifier: to make a small signal larger, nothing more or less.

This GaN amp, with it's lower power, higher distortion, and lower switching frequency, doesn't improve upon the state of the art class d such as Purifi in any material way, despite a price 3x the Purifi, much less meaningfully exploit the inherent advantages of GaN technology. It might appeal to a certain group of users who like higher distortion of a certain type, who don't need 400 watts, and who might like less heat, more reliability, and a smaller form factor. In other words, the traditional tube loving customers of Atmosphere.

Regarding distortion audibility, a designer of any amp has two courses: either get it so low that it is actually and for real masked by the music (which IMO/IME is lower than traditionally though; well below -100dB on account of the ear using the higher orders to sense sound pressure and has a pretty wide range; it is more sensitive to them than almost anything else) or mask the distortion by using innocuous lower orders (which is how tube amps do it). The problem with the tube amp way is the 2nd in particular causes a bit of 'warmth' (that's the tonality the ear assigns and is not a FR error) which some find 'objectionable' although when challenged find its pretty hard to point out. Personally I find 'bright and harsh' objectionable (and is easy to point out); its as much a coloration as 'warm' and unlike the lower orders, humans in general find them objectionable.

We didn't design this circuit for any particular distortion signature- the 'chips fell where they may' if you see what I did there . The lower orders turn up in some class D designs on account of dead time and non-linearities in the encoding scheme. When the amp was 'done' (meaning that we were meeting EU Directives for radiation) we found it sounded a lot like our OTLs but even more transparent owing to vastly reduced distortion and being a nearly perfect voltage source. This was a pleasant surprise since in the world of tubes, transparency and lack of colorations has kept us in business the last half century or so. This module is capable of 400 watts but we wanted it to manage its heat well. It can sit at 200 watts into 4 Ohms all day long and not break a sweat.

 

[SIY]

Thanks, regarding the former, the 'evidence' is all around us. I would start with the Radiotron Designer's Handbook (I looked this one up when John Curl mentioned it for the same reason I do) 3rd edition, published in the 1930s. In it, fairly early on you see that there was an understanding at that time that the ear was considerably more sensitive to the higher ordered harmonics than the lower orders.

That simple fact is what has kept tubes in business all these years. That tubes are still around is the second bit of evidence.

What is less understood is that the 2nd and 3rd (both of which are treated by the ear the same way, which is to say they are innocuous) can mask the higher ordered harmonics, similar to how this happens in a loudspeaker. So a tube amp, which generates considerably more higher ordered harmonics than most any solid state amp, won't sound bright and harsh on that account because the harmonics are masked. If its not clear at this point it should be: the ear/brain system assigns a tonality to all forms of distortion and it is most sensitive to the higher orders since it uses them to sense sound pressure.




Actually the amp was built because I came to the opinion about 5 years ago that if a manufacturer of amps in the high end audio market didn't figure out class D pretty soon, they were going to get left behind; this is gradually playing out in the high end audio market. We might be a little weird compared to a number of high end audio companies in that we actually have engineering capacity in house, using someone else's modules seemed to send the wrong message. We were issued a patent about a year and a half ago.



Regarding distortion audibility, a designer of any amp has two courses: either get it so low that it is actually and for real masked by the music (which IMO/IME is lower than traditionally though; well below -100dB on account of the ear using the higher orders to sense sound pressure and has a pretty wide range; it is more sensitive to them than almost anything else) or mask the distortion by using innocuous lower orders (which is how tube amps do it). The problem with the tube amp way is the 2nd in particular causes a bit of 'warmth' (that's the tonality the ear assigns and is not a FR error) which some find 'objectionable' although when challenged find its pretty hard to point out. Personally I find 'bright and harsh' objectionable (and is easy to point out); its as much a coloration as 'warm' and unlike the lower orders, humans in general find them objectionable.

We didn't design this circuit for any particular distortion signature- the 'chips fell where they may' if you see what I did there . The lower orders turn up in some class D designs on account of dead time and non-linearities in the encoding scheme. When the amp was 'done' (meaning that we were meeting EU Directives for radiation) we found it sounded a lot like our OTLs but even more transparent owing to vastly reduced distortion and being a nearly perfect voltage source. This was a pleasant surprise since in the world of tubes, transparency and lack of colorations has kept us in business the last half century or so. This module is capable of 400 watts but we wanted it to manage its heat well. It can sit at 200 watts into 4 Ohms all day long and not break a sweat.

Again. what's missing is any evidence that your amps sound different than low distortion designs once impedance-related frequency response errors are accounted for. You've espoused the "distortion profile" hypothesis (as famously has Nelson), but at the levels your amps produce (relatively low compared to amps that are designed rather than engineered), the hypothesis begs for evidence, not just that it explains sonic differences but that those differences (frequency response aside) even exist.

So far, nope. I understand, you have your market and they don't care, so why bother? But hey, I'm a (now former) scientist, so the whole hypothesis testing thing is hard-wired into me.

 

[antcollinet]

...... That tubes are still around is the second bit of evidence......

Gonna take you up on this. Crystals to place on speaker cables are still around - as are cable lifters, as are $10,000 power leads, as are liquids to paint on your connector terminals, as are all sorts of other audio tomfoolery.

That is not evidence that any of them offer any benefits at all.

 

[AudioJester]

Fixed hardware distortion/harmonic "tuning" -whether tube or class D seems very old school, rigid and lacking flexibility.
Something like E1DA dongle dac and android app with ability to set 2nd and 3rd harmonic levels is surely a better way to move forward. Then each user can adjust to their own preference, and experiement as they wish without needing to change hardware. With the right gear you can also chose/set your own frequency response / curve etc.

 

[goryu]

As a start, it would be nice to see a full complement of measurements to actually see the distortion spectra, frequency response with load, etc.

It is my understanding that class d amps require a highly sophisticated knowledge base and skill set to overcome a steep learning curve as they present a whole different set of challenges in comparison to traditional class a and ab amps, both ss and tubes. From what Bruno Putzeys has said, they do everything but want to work, so kudos to an old tube amp designer for tackling the class d conundrum and succeeding in getting a product to market.

 

[atmasphere]

Again. what's missing is any evidence that your amps sound different than low distortion designs once impedance-related frequency response errors are accounted for.

Ah. It was not clear to me in the past that this is what you were looking for. I have to assume that impedance-related FR isn't a thing since any amplifier that makes low distortion will also have a low output impedance so should be quite flat. In our case its difficult to measure, being about 10milliOhms. So in this regard, no, I don't have any evidence as we've yet to conduct any studies in that regard. Nor do I think there will be much difference, FWIW. I don't see this as me somehow living in a different universe

Gonna take you up on this. Crystals to place on speaker cables are still around - as are cable lifters, as are $10,000 power leads, as are liquids to paint on your connector terminals, as are all sorts of other audio tomfoolery.

That is not evidence that any of them offer any benefits at all.

I agree! FWIW though tubes are a bit different from magic stones and rain forest wood cable lifters in that they were the dominant form of amplification. At no time in the past did intelligent chips placed on a CD player or tiny little 'dots' form any sort of basis for amplification, or any other such tweaks.

Usually when a new technology appears (such as transistors) the prior art is relegated to the dust bins unless collectors show up to keep them alive. For example there are no side valve motors made in the US but at one time they were the only type of valve used in internal combustion (although cars and motorbikes with side valve motors can be quite collectable...). The overhead valve supplanted them as the increased power was easy to discern. Tubes were declared 'obsolete' sometime in the 1960s (when germanium was king) yet somehow they continue to this day. They've actually been in production longer now than when they were declared obsolete.

The reason this is so is that to get the Gain Bandwidth Product to really allow feedback to do its job properly (i.e. distortion does not rise with frequency) you need lots of open loop gain and lots of bandwidth to go with it and the semiconductors needed to do that didn't exist in the 1960s and well into the 1990s (opamps being a bit of a variable exception but difficult to integrate into a power amplifier design without adding lots of out of passband poles). I avoided solid state when I was younger simply because the amps at the time sounded harsh at volume. Many current designs still do. Bruno showed that class D offers a way around that (by being able to run enough feedback); if you've not seen this article its worth a read:
https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf

You'll notice he says the same thing I do about the prior art; pay special attention to pages 14-16.

 

[atmasphere]

From what Bruno Putzeys has said, they do everything but want to work,

I can confirm that

 

[antcollinet]

Ah. It was not clear to me in the past that this is what you were looking for. I have to assume that impedance-related FR isn't a thing since any amplifier that makes low distortion will also have a low output impedance so should be quite flat. In our case its difficult to measure, being about 10milliOhms. So in this regard, no, I don't have any evidence as we've yet to conduct any studies in that regard. Nor do I think there will be much difference, FWIW. I don't see this as me somehow living in a different universe

I agree! FWIW though tubes are a bit different from magic stones and rain forest wood cable lifters in that they were the dominant form of amplification. At no time in the past did intelligent chips placed on a CD player or tiny little 'dots' form any sort of basis for amplification, or any other such tweaks.

Usually when a new technology appears (such as transistors) the prior art is relegated to the dust bins unless collectors show up to keep them alive. For example there are no side valve motors made in the US but at one time they were the only type of valve used in internal combustion (although cars and motorbikes with side valve motors can be quite collectable...). The overhead valve supplanted them as the increased power was easy to discern. Tubes were declared 'obsolete' sometime in the 1960s (when germanium was king) yet somehow they continue to this day. They've actually been in production longer now than when they were declared obsolete.

The reason this is so is that to get the Gain Bandwidth Product to really allow feedback to do its job properly (i.e. distortion does not rise with frequency) you need lots of open loop gain and lots of bandwidth to go with it and the semiconductors needed to do that didn't exist in the 1960s and well into the 1990s (opamps being a bit of a variable exception but difficult to integrate into a power amplifier design without adding lots of out of passband poles). I avoided solid state when I was younger simply because the amps at the time sounded harsh at volume. Many current designs still do. Bruno showed that class D offers a way around that (by being able to run enough feedback); if you've not seen this article its worth a read:
https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf

You'll notice he says the same thing I do about the prior art; pay special attention to pages 14-16.

Gonna defer to your almost infinitely greater experience with amplifier design, and just accept that there may be reasons for tube designs to exist. However my point wasn't that tubes offer no benefit, it was that (in a world where millions believe that all sort of woo will improve the sound from their systems); the fact that they're still around offers little in the way of evidence for that benefit.

 

[atmasphere]

Gonna defer to your almost infinitely greater experience with amplifier design, and just accept that there may be reasons for tube designs to exist. However my point wasn't that tubes offer no benefit, it was that (in a world where millions believe that all sort of woo will improve the sound from their systems); the fact that they're still around offers little in the way of evidence for that benefit.

I get it. I don't buy into the woo thing- I find it annoying. 'Tweaks' have to have measurable benefit, one that is measurable and audible. A tweak also had better have readily discernible reasons why it should work for an engineering mind- some cost quite a lot

 

[BILGUANA]

I have found that the Orchard Ultimate Stereo amp sounds much better than VTV Purifi monoblocks. Purifi seems to be stuck in the past.

 

[SIY]

I have found that the Orchard Ultimate Stereo amp sounds much better than VTV Purifi monoblocks. Purifi seems to be stuck in the past.

Uh huh.

 

[VMAT4]

Wait until the next two issues of AudioXpress.

Oooooooo! Can hardly wait. Of course I bought an Hypex based Apollon Amp.

 

[goryu]

I have found that the Orchard Ultimate Stereo amp sounds much better than VTV Purifi monoblocks. Purifi seems to be stuck in the past.

lol...My understanding is the Orchard amp uses the flavor of the day GaN devices, which really don't offer any real improvements/advantages when used in class d amps at the lower switching rates (500-800kHz) most are using. The circuit topographies in all of these GaN amps are not to my knowledge as sophisticated/state of the art as that used in the Purifi.

 

[atmasphere]

I wonder where this is going? Jim

No idea... Purifi stuck in the past?? IMO pretty much defines the future...

 

[goryu]

maybe Mike is back, lol...

 

[JustJones]

@atmasphere

Any interest in sending in a pair for measurements?

 

[atmasphere]

@atmasphere

Any interest in sending in a pair for measurements?

In due time, sure. We're just trying catch up with demand at the moment.

 

[Dweaton]

Ive had numerous class d amps in my controlled listening room over the years.
Unlike most audiophiles here I work on music professionally and have a fully treated room. My speakers are set up in a perfect equilateral triangle and my tweeters are ear level. My phantom center is insanely good.

My speakers are modified Immedia Allegra that under went a full 9k crossover rebuild with cost no object parts. I use Benchmark, Bricasti, and Mytek conversion here.

My room: Not a few bass traps and thin panels like most of the consumers here but a full array of treatment to flatten out my room. My speakers are decoupled and overall it is a very hyper critical listening room.

(Btw I find it humorous how spec driven this website is even though I find it important to me it’s only part of story as system synergy, room treatment, and low level listening factor in) One of my favorite sounding AD converters is a JCF latte which doesn’t have the best measurements but has a really great/clean sound in analogue to digital capture that feels so musical and right) also worth looking into Playback Designs DACs and ADCs and what the designer there writes about specs…)

I guess I’m saying I appreciate this website but much of it I take for a grain of salt as most of the users here have untreated rooms. (Putting a few panels up and doing a system EQ is not a treated room..) I love some
The findings here though and see the merit overall.

Non-the less these minimal specs measurements done here don’t always equate to everything for me. For instance I enjoyed the new purifi modules in the nad c298 amp recently had it for 3months but felt at low volumes the music didn’t sustain) hard to describe) fyi I could live with them and work with them as they are very neutral/clean)…
these amps definably made me re-think class-d as every other amp I’ve used that has been class-d ended up annoying me sonically after time… almost like I was hearing a box of moving parts rather than music…

I’ve demoed Orchard amps, Wyred4sound (ice based modules) not my cup of tea…, and recently th e new AGD mono amps. (GaN) weirdly when we were switching between both the Orchards and the AGD monos. There was a strange fatiguing quality to the Orchards compared to the AGDs. Like in some chorale music we were listening to in the upper midrange it felt a little hard and grainy on the ears (note: the orchards are amazing overall and super fluid/fast and overall they really blew me away at first) and I think if we didn’t have the AGD amps we wouldn’t have noticed it. (This was even experienced in another room with other speakers through a high end MSB DAC)

All in all these new GAN amps really have a subtle different sound to them than the purifi based amps) it’s really like no sound almost like straight wire with gain… there is a fluidity and natural presentation to the music. I’m sold on them and thinking to switch now to using either a pair of AGD or the maxed out Orchards.
I’m not bothered by the specs of this amp…
Yet here we talk about specs of GAN amps and purifi modules… if you demo and listen to amps as much me there is clearly a different “signature” to these amp topologies… the sound I prefer is from the GAN amps… I know this is not a “feeling” forum. But it feels more like music… lol
Maybe I’m the only one here curious to demo these Atma amps…

Note: recently a few months ago I went to a local audiophiles house (70k amps, 20k DAC, and the whole 9 yards…) his room was bright and super fatiguing… no acoustics and really a painful experience overall. I was polite when I was there… long story short this guy is a big reviewer on audiogon… people send him equipment to review, he doesn’t even level match amps/dacs when he A/Bs by writes really flowery reviews of gear (not to mention R2R DAC promoter… I was disgusted… I appreciate this website a lot as there are many people like that spewing nonsense.

One the other side——> a month before that I went to an audiophiles house in Malibu… older man with endless money. Pure silver wired system(I looked up the speaker cables and they were 90k each), 300k audio note amps, and million dollar plasma coil speakers through horn loaded drivers, additional subs (forgot the models)… basically a 2-3 million dollar system… this system was amazing. 4 watts amps through 118db sensitivity speakers. I’ve never heard music float like that. Was it accurate? I don’t care. It blew me away. Music is wiggles in the atmosphere…