Review: Atma-Sphere Class D Amplifiers

[Atmasphere]

Of course you did. You chose the parts and you chose the circuit according to what you count as good.

If part specification alone is how we count as 'good' then this is correct.

Every time I've seen anyone talk about 'voicing' though, its been in the context of choosing certain parts to obtain a tonality. That is something we didn't do. So this, for me, is a new definition of the word 'voicing' of which I was not previously aware

IOW quite literally when we had all the parts together, when the last issue of switching noise was solved, we had a functioning amp that went to market. At no time did we ever audition the 'sound' of a certain part in the design.

When doing the design work on any of our products we have based the part choice on part specification. In the tube stuff, which uses coupling caps, the parts choice was based on capacitor specification rather than what people were saying about the parts or what we heard. But since zero feedback circuits are so sensitive to parts and literally everything, we could hear differences. The differences were always heard after the fact and I can't think of a time when we tried a part with good specs that we didn't wind up using.

 

[Atmasphere]

Edit: I understand voicing as choosing the components (transformers, resistors, etc.) and the character of the circuitry to achieve a particular sonic result.

I've been thinking about this since you mentioned it.

While so far I don't think we did that for the class D, for our tube circuits I have always designed for a particular result: when the amp or circuit is overloaded (clips) I want to see symmetrical distortion, which is going to result in a 3rd harmonic being the dominant distortion component. That is done by selecting the correct operating point. I often find that near the circuit's maximum gain.

Since the distortion signature is literally the 'sound' of most amplifiers or preamps, unless there is a significant deviation from flat frequency response, the result will be a relaxed character in the presentation, since the 3rd harmonic is treated much like the 2nd by the ear in that it imparts a sense of 'warmth'.

That is why tape recorders have a following, as any analog recorder if working properly will have a 3rd harmonic as its dominant distortion component as the tape nears saturation.

So in that way I have been 'voicing' our products; I just didn't think about it that way as I was intent on the distortion signature rather than some sort of frequency response variation. Speaker designers often have to sculpt their crossovers or have compensation to deal with pesky characteristics of the drivers they are using in order to get their speaker to sound right. I was thinking of 'voicing' in that manner and so was probably being a bit myopic.

 

[MarkusBarkus]

...thank you for that explanation.

 

[PeterA]

Interesting that Tim was able to draw that out.

 

[tima]

Since the distortion signature is literally the 'sound' of most amplifiers or preamps, unless there is a significant deviation from flat frequency response, the result will be a relaxed character in the presentation, since the 3rd harmonic is treated much like the 2nd by the ear in that it imparts a sense of 'warmth'.

Interesting comments, Ralph, thanks.

I want to draw out a point I hinted at upstream that speaks to voicing in a different way.

From my experience I"ve found each amplifier or preamplifier I've heard sounds different to each other amp or preamp. I've found each amp or preamp from the same manufacturer sounds much less different than other models from the that manufacturer than they do sound different from those of a different manufacturer.

That seems to make sense. If the same designer/engineer designs and builds, say, three tiers of two models each (amp and preamp) there will be many similarities across the tiers with sufficient difference to set apart the tiers from one another. However many tiers doesn't really matter. That there are differences is intentional.

That is what I"ve experienced from the manufacturers of gear I've owned or reviewed that offers tiers, such as Conrad-Johnson, Audio Research, Lamm and to a more limited externt Atma-Sphere. The ARC Ref5SE linestage sounds different from but very simlar to the ARC REF10 linestage. The Atmasphere M-60 sounds different from the Atmaspher MA-1 but they are also very close sonically. And likewise the 140W MA-1 sounds quite different from the 140W Conrad-Johnson Premier 140.

There is what I might call a heritage of linearity across manufacturers models. Each brand has its own voice. That is not by accident. I don't know if it goes too far to say the designer/engineer and the methods he/they use for choosing components and circuits are the unique voicing the brand exhibits. While methods for discovering improvement may vary over time, the brand remains true to its voice because the designer/engineer is its voice.

In the few cases where a brand is successfully passed on to a new designer/engineer (eg. ARC, Wilson), that voice, the voice of the original owner carries across the transition.

 

[Mgrif104]

Interesting comments, Ralph, thanks.

I want to draw out a point I hinted at upstream that speaks to voicing in a different way.

From my experience I"ve found each amplifier or preamplifier I've heard sounds different to each other amp or preamp. I've found each amp or preamp from the same manufacturer sounds much less different than other models from the that manufacturer than they do sound different from those of a different manufacturer.

That seems to make sense. If the same designer/engineer designs and builds, say, three tiers of two models each (amp and preamp) there will be many similarities across the tiers with sufficient difference to set apart the tiers from one another. However many tiers doesn't really matter. That there are differences is intentional.

That is what I"ve experienced from the manufacturers of gear I've owned or reviewed that offers tiers, such as Conrad-Johnson, Audio Research, Lamm and to a more limited externt Atma-Sphere. The ARC Ref5SE linestage sounds different from but very simlar to the ARC REF10 linestage. The Atmasphere M-60 sounds different from the Atmaspher MA-1 but they are also very close sonically. And likewise the 140W MA-1 sounds quite different from the 140W Conrad-Johnson Premier 140.

There is what I might call a heritage of linearity across manufacturers models. Each brand has its own voice. That is not by accident. I don't know if it goes too far to say the designer/engineer and the methods he/they use for choosing components and circuits are the unique voicing the brand exhibits. While methods for discovering improvement may vary over time, the brand remains true to its voice because the designer/engineer is its voice.

In the few cases where a brand is successfully passed on to a new designer/engineer (eg. ARC, Wilson), that voice, the voice of the original owner carries across the transition.

Interesting insight Tim. While I had experienced what you are suggesting might be called a heritage of linearity, I hadn’t really thought much about it before - other than how it influenced and/or reinforced my own biases for or against certain brands. I appreciate the insight.

 

[Atmasphere]

Interesting comments, Ralph, thanks.

I want to draw out a point I hinted at upstream that speaks to voicing in a different way.

From my experience I"ve found each amplifier or preamplifier I've heard sounds different to each other amp or preamp. I've found each amp or preamp from the same manufacturer sounds much less different than other models from the that manufacturer than they do sound different from those of a different manufacturer.

That seems to make sense. If the same designer/engineer designs and builds, say, three tiers of two models each (amp and preamp) there will be many similarities across the tiers with sufficient difference to set apart the tiers from one another. However many tiers doesn't really matter. That there are differences is intentional.

That is what I"ve experienced from the manufacturers of gear I've owned or reviewed that offers tiers, such as Conrad-Johnson, Audio Research, Lamm and to a more limited externt Atma-Sphere. The ARC Ref5SE linestage sounds different from but very simlar to the ARC REF10 linestage. The Atmasphere M-60 sounds different from the Atmaspher MA-1 but they are also very close sonically. And likewise the 140W MA-1 sounds quite different from the 140W Conrad-Johnson Premier 140.

There is what I might call a heritage of linearity across manufacturers models. Each brand has its own voice. That is not by accident. I don't know if it goes too far to say the designer/engineer and the methods he/they use for choosing components and circuits are the unique voicing the brand exhibits. While methods for discovering improvement may vary over time, the brand remains true to its voice because the designer/engineer is its voice.

In the few cases where a brand is successfully passed on to a new designer/engineer (eg. ARC, Wilson), that voice, the voice of the original owner carries across the transition.

Since distortion literally distinguishes the different 'house sound', its pretty obvious that the design approach is governing the result.

IMO/IME the resulting 'sound' from any design should be relaxed and detailed at the same time so the most resolution is obtained in a musical way that allows one to listen all day; IOW is engaging and not fatiguing.

 

[PeterA]

IMO/IME the resulting 'sound' from any design should be relaxed and detailed at the same time so the most resolution is obtained in a musical way that allows one to listen all day; IOW is engaging and not fatiguing.

I totally agree. My favorite systems have all shared these characteristics.

 

[microstrip]

Since distortion literally distinguishes the different 'house sound', its pretty obvious that the design approach is governing the result.

Probably this distortion pattern is mostly due to circuit topology and the operating points of tubes. IMO, although keeping their sound signature. the more iconic brands evolved towards a more neutral sound, reducing their sound signatures along time.

IMO/IME the resulting 'sound' from any design should be relaxed and detailed at the same time so the most resolution is obtained in a musical way that allows one to listen all day; IOW is engaging and not fatiguing.

All gear designers claim such epithets. BTW, resolution is not just "more", we can have different types of resolution.

 

[Atmasphere]

Probably this distortion pattern is mostly due to circuit topology and the operating points of tubes.

Yes; this is so for transistors as well including class D.

Some examples:
A fully differential topology will exhibit a cubic non-linearity. So the third harmonic is dominant with succeeding harmonics falling off on an exponential curve. The third will be at or slightly less than the level seen in a single-ended circuit, which expresses a quadratic non-linearity. That will have the 2nd harmonic as dominant. The succeeding harmonics also fall off on an exponential curve as the order of the harmonic is increased, but owing to the exponent, do not fall off as rapidly as seen with a cubic non-linearity. So the fully differential circuit is inherently lower distortion.

This does not matter solid state or tube. Mixing the two topologies in the same circuit results in algebraic summing of the two non-linearities. The 5th harmonic often gets some emphasis on this account.

Feedback complicates things. In 99% of amps made, feedback is applied to a non-linear node near the input of the amp. Depending on the nature of the non-linearity, the feedback signal is distorted in various ways, resulting in a spray of harmonics and inharmonic (intermodulation) information not native to the original circuit operating open loop. On this account I'm of the opinion that feedback can only be mixed in a resistive divider network in a manner identical to how its done with opamps. This causes the feedback to have less deleterious effects. This is the tip of the iceberg...

 

[tima]

Funny what I keep discovering amongst my piles of videos ... I need better organizing.

Here are the Class D amps doing the Soria series Reiner Verdi (RCA LDS-6091) that Ralph has mentioned for its huge dynamic range. A bit tricky for recording in terms of setting the volume with parts ranging from pppp pianissississimo to ffff fortissississimo.

I believe this is some of the Requiem (mvmnt 1) and some of the Dies Irae (mvmnt 2). I wonder how many choristers in the choir. Iirc it requires a double choir which means 100+ at least. Plus the soloists.

If you want a sense of the large end of that, start around 5:30 and play at least through say, 8:00 but the whole thing is worth a listen. At least I like it The Class Ds acquit themselves very well -- this is quite a test for audio reproduction.

RCA LDS-6091

 

[Atmasphere]

A bit tricky for recording in terms of setting the volume with parts ranging from pppp pianissississimo to ffff fortissississimo.

Essentially the percussionist has to hit the drum heads as hard as he can!

 

[MarkusBarkus]

...well those very groovy Class D monos arrived midday, and I finally hooked things up, sort-of temporarily, as I have a small project going on that will require a different arrangement.

But, right out of the box, they sound excellent. I am puttin' the pot on and cookin' over here this afternoon. Very tasty.

Upstairs system:
--Atma-Sphere Class D monos.
--Deconstructed Chord DAVE, pre and dac source.
--Denefrips D2D
--Aurender ACS10 streamer.
--Magico A5 speakers.

 

[tima]

...well those very groovy Class D monos arrived midday, and I finally hooked things up, sort-of temporarily, as I have a small project going on that will require a different arrangement.

But, right out of the box, they sound excellent. I am puttin' the pot on and cookin' over here this afternoon. Very tasty.

Upstairs system:
--Atma-Sphere Class D monos.
--Deconstructed Chord DAVE, pre and dac source.
--Denefrips D2D
--Aurender ACS10 streamer.
--Magico A5 speakers.

I'll be interested to learn your assessment after several hundred hours. I found the bass is the last to come in. Have fun!

 

[MarkusBarkus]

...having read and enjoyed your review, I began with the bassiest of bass tracks. And it was very good from note-one. If things get better, well, that's even mo' better.