Atma-Sphere Class D Mono blocks

[analogsa]

Ralph, I know you have spoken many tines about this. If you are feeling generous, would you remind me ( or link to prior discussion ) a bit more in detail about the rise of distortion into decreasing impedance? For example is it mostly IMD etc.?

This is interesting to me as it seems many makers of class D amps encourage lower impedance loads ( especially in pro products ) as their amps 'make more power' into those loads.

Thanks in advance...

Not Ralph here, but also find the topic interesting.

Non switching amps at least, are very sensitive to being loaded. If you measure the unloaded voltage output of practically any amp you would be amazed at how low the distortion is. Once current starts flowing though, the picture changes, often dramatically. It is mainly due to non linearities in the transfer characteristics of the output devices, but even the voltage amplification stages may be sensitive to the output current.

Look at Stereophile measurements, every time the load halves, distortion approximately doubles. Irrespective of class A/B o the number of output devices. With all else being equal an 8ohm speaker causes less amplifier harmonic distortion than a 4 ohm.

And IMD is always the consequence of harmonic distortion.

 

[Solypsa]

Got it. So why the push towards low impedance? Is it a case of the overall distortion level being low enough that engineers think " it's so low it doesn't matter if it doubles or quadruples"?

 

[Hear Here]

With all else being equal an 8ohm speaker causes less amplifier harmonic distortion than a 4 ohm.

Got it. So why the push towards low impedance?

In my early days of hi-fi, most speakers were 15 ohm and most amps were valve jobs. Then, when ss took over, speaker impedance often changed to 8 or even 4 ohms. Why should they do this if it increases distortion? Don't ask me, unless the extra distortion (if in fact it really does exist) actually improves the listening experience.

As it happens my speaker brand of choice recently (a few years ago now) moved from 6 ohm nominal to 18 ohms. Was this an effort to reduce distortion or to make their speakers even more valve-friendly? Don't ask me, but they certainly sound good with Class D behind them - or valves for that matter.

 

[D.Duttilleux]

Thought WBF was a politics free zone?

Anyway, who cares about politics when you can enjoy the glorious symphony of a 90° L-Twin (270°- 450º firing sequence) with desmodromic valve system, emitted through Termignoni Corsa pipes?

3 years ago (note the time frame), I had a conversation with a manufacturer who makes systems that - like Atma-Sphere and Classic Audio - are inspired by historical know-how. We then concluded that audio can serve as a metaphor for 'everything'.

 

[Atmasphere]

Ralph, I know you have spoken many tines about this. If you are feeling generous, would you remind me ( or link to prior discussion ) a bit more in detail about the rise of distortion into decreasing impedance? For example is it mostly IMD etc.?

This is interesting to me as it seems many makers of class D amps encourage lower impedance loads ( especially in pro products ) as their amps 'make more power' into those loads.

Thanks in advance...

IMD and harmonic distortion increase with reduced load impedance. There is nothing anyone can do about it except get a speaker that has a higher impedance.

Got it. So why the push towards low impedance? Is it a case of the overall distortion level being low enough that engineers think " it's so low it doesn't matter if it doubles or quadruples"?

High efficiency speakers are expensive to build. But tube power was even more expensive, so in the old days speakers that made the most of tube power were popular. Then along came cheaper transistor power: no output transformer and no filament supply. Transistor amp manufacturers found they could charge about 90% of what the same amp in a tube embodiment cost so they made more money. As time went by transistor amps got cheaper as better semiconductors at lower prices showed up. Meanwhile, people figured out that if the solid state amp was built properly, you could double the output power into half the impedance. Since less efficient speakers are so much cheaper to build, speaker manufacturers saw they could add 3dB to their (sensitivity) figures by going 4 Ohm rather than 8. Its a horse race of sorts, in pursuit of the almighty dollar than anything to do with performance for the human ear.

The 'slight' increase in distortion was considered 'negligible'. For me, although sometimes the use of that word is justified, it always raises a red flag. Distortion is rarely inaudible; its a modifier of the signal that passes through the system and so imposes a tonality on the signal, usually easily heard. We've always paid attention to this fact; it makes a difference in how you design a product.

IMO there's no good excuse for low impedance speakers unless the only way to do the technology of the speaker is a low impedance embodiment. But speaker manufacturers don't see it that way; IMO simply because they don't always understand the relationship between the amp and speaker and how audible distortion really is.

So even though our class D easily doubles power as the load impedance is halved (right up to its current limit protection) I'm of the opinion that the easier the speaker is to drive the better, especially since the class D has a really nice first Watt.

 

[Hear Here]

So even though our class D easily doubles power as the load impedance is halved (right up to its current limit protection) I'm of the opinion that the easier the speaker is to drive the better, especially since the class D has a really nice first Watt.

Interesting and I'm happy to hear what you say about speaker impedance and sensitivity.

My speakers are 18 ohm and 107 dB. With your Class D amps, I normally limit the preamp output to -12dB for speaker and ear protection (and neighbour consideration), but if I allow full output, I find that a loud but not over-powering volume (around 80 dB at listening position measured by my phone's grotty Sound Meter) using your amps is around -35 dB. I'm not sure how many watts I'm using, but there are plenty more in reserve!

 

[D.Duttilleux]

IMD and harmonic distortion increase with reduced load impedance. There is nothing anyone can do about it except get a speaker that has a higher impedance.

High efficiency speakers are expensive to build. But tube power was even more expensive, so in the old days speakers that made the most of tube power were popular. Then along came cheaper transistor power: no output transformer and no filament supply. Transistor amp manufacturers found they could charge about 90% of what the same amp in a tube embodiment cost so they made more money. As time went by transistor amps got cheaper as better semiconductors at lower prices showed up. Meanwhile, people figured out that if the solid state amp was built properly, you could double the output power into half the impedance. Since less efficient speakers are so much cheaper to build, speaker manufacturers saw they could add 3dB to their (sensitivity) figures by going 4 Ohm rather than 8. Its a horse race of sorts, in pursuit of the almighty dollar than anything to do with performance for the human ear.

The 'slight' increase in distortion was considered 'negligible'. For me, although sometimes the use of that word is justified, it always raises a red flag. Distortion is rarely inaudible; its a modifier of the signal that passes through the system and so imposes a tonality on the signal, usually easily heard. We've always paid attention to this fact; it makes a difference in how you design a product.

IMO there's no good excuse for low impedance speakers unless the only way to do the technology of the speaker is a low impedance embodiment. But speaker manufacturers don't see it that way; IMO simply because they don't always understand the relationship between the amp and speaker and how audible distortion really is.

So even though our class D easily doubles power as the load impedance is halved (right up to its current limit protection) I'm of the opinion that the easier the speaker is to drive the better, especially since the class D has a really nice first Watt.

Although 16 Ohm drivers are still widely available from the major pro audio brands, the construction of modern drivers is not 1 to 1 comparable to the classic RCA, Western Electric and early Lansing woofers. The diameter of the voice coils was on average much smaller and the tolerances were narrower, resulting in, among other things, a much higher flux density. Xmax was hardly relevant, because woofers weren't yet used as 'air pump'.


Mr Karstens, I would like to ask you a question that arises from discussions with my brother about GaN FET Class D amplifier tech.
Would it make sense to manufacture the mono blocks with high quality 'discrete' through-hole parts instead of surface mounted PCBs?

 

[Atmasphere]

Would it make sense to manufacture the mono blocks with high quality 'discrete' through-hole parts instead of surface mounted PCBs?

It is possible to use through-hole parts but you'll limit the switching frequency and more advanced parts like GaNFETs couldn't be used. You might even get into trouble using MOSFETs. Through-hole (leaded) parts have inductance in their leads, which easily results in parasitics that can be considerably higher than the switching frequency of the amp. You also have greater noise due to longer traces and the holes in the board. Plus multiple layers in the circuit board are much harder to execute!

Once you understand how to work with Surface Mount Devices they really aren't that bad to work with, although you sometimes have to be careful not to sneeze if you want to keep all your parts in front of you You use different tools and one of them is a set of magnifying glasses so you can see what you're doing. All of our boards are thus built by hand.

 

[D.Duttilleux]

It is possible to use through-hole parts but you'll limit the switching frequency and more advanced parts like GaNFETs couldn't be used. You might even get into trouble using MOSFETs. Through-hole (leaded) parts have inductance in their leads, which easily results in parasitics that can be considerably higher than the switching frequency of the amp. You also have greater noise due to longer traces and the holes in the board. Plus multiple layers in the circuit board are much harder to execute!

Thank you for your answer, which reflects my arguments from the conversation with my brother.
He is convinced of the importance of high-quality (discrete) components, but ignores the impact of parasitics in high-frequency circuits.
Ultimately it's about Electromagnetic Compatibility (EMC).

 

[Nuforce]

View attachment 118934
Cold off the FedEx truck with about 15 minutes of mains time. They sound very detailed and balanced. Wow, when these things get thoroughly warm and settled we know they will be better.

Looks like you have identical amps to me, older style MA-1?. How are you finding the class D amps?

 

[DetroitVinylRob]

Looks like you have identical amps to me, older style MA-1?. How are you finding the class D amps?

Yes, MA-1 Silver Editions, fully updated. Absolutely ??these amps!

I’ve been away on the gulf coast for a while, just looking forward to getting back into the kit and some quality listening. Had a few brief and enjoyable sessions with friends and some back and forward with the MA-1s. Some find it an either or, I’m not there (yet).

My first impressions is simply how much they sound like the OTLs. Very easy to listen to, very articulate, and offer a well balanced musical experience. They are different though in the way they present. Perhaps the mid bass is a little fuller, the D class is dead quiet when no music is playing, and they are on point right after you flip the switch, I don’t believe there is much change in sound at all as time goes by. I would say “as they warm up”, but they simply don’t, and for me, the chassis never feel above room temperature. Amazingly, no change in sound quality.

Ralph and his team have really accomplished something quite remarkable here.

More listening…

 

[iansr]

“My first impressions is simply how much they sound like the OTLs.”
One common theory as to why class A amps sound different to other amps is that their harmonic distortion profile is different, specifically that 2nd order HD is dominant and the higher orders decrease in magnitude as you go up the orders. Most class D amps are designed to minimise all HD. So I’m wondering what it is about Ralph’s class D amps that make them sound like his OTLs. Perhaps Ralph would care to comment ?

 

[morricab]

I’ve been away on the gulf coast for a while, just looking forward to getting back into the kit and some quality listening. Had a few brief and enjoyable sessions with friends and some back and forward with the MA-1s. Some find it an either or, I’m not there (yet).

My first impressions is simply how much they sound like the OTLs. Very easy to listen to, very articulate, and offer a well balanced musical experience. They are different though in the way they present. Perhaps the mid bass is a little fuller, the D class is dead quiet when no music is playing, and they are on point right after you flip the switch, I don’t believe there is much change in sound at all as time goes by. I would say “as they warm up”, but they simply don’t, and for me, the chassis never feel above room temperature. Amazingly, no change in sound quality.

Ralph and his team have really accomplished something quite remarkable here.

More listening…

What do you mean by "They are different in how they present"?

 

[Nuforce]

I am very attached to my MA-1's however I might consider trying a pair of the class D - I find the need for AC in summer a bit wasteful in terms of resources.

 

[Atmasphere]

One common theory as to why class A amps sound different to other amps is that their harmonic distortion profile is different, specifically that 2nd order HD is dominant and the higher orders decrease in magnitude as you go up the orders.

If that is a theory, here is will be debunked.

The distortion profile of any amplifier is more dependent on its topology than its class of operation. For example, if the class A amplifier is fully differential and balanced, the dominant distortion product will be the 3rd, not the 2nd, even if its a triode amplifier. This is because of distortion cancellation from stage to stage in the circuit: its topology dependent. If you change the class A operation to AB, the 3rd remains the dominant distortion product.

The traditional problem with solid state has been brightness and harshness caused by higher ordered harmonics. Those harmonics are often caused by a misapplication of feedback; if there is insufficient Gain Bandwidth Product to support the gain of the amp and the amount of feedback used, distortion can climb above a certain frequency as the distortion falls off at the GBP limit. Its very easy with class D to generate very high values of GBP (very high loop gain) so you can prevent this particular phenomena happening in the audio band.

Ideally you want the distortion vs frequency to be ruler flat line across the audio band. Zero feedback tube amps can do this fairly easily. I've become convinced that this is one of the more telling aspects of how the amplifier is going to 'sound'; that you can draw a direct correlation between the measurements and sound quality starting with this measurement.

The distortion profile is the other big thing- it really must be benign to the human ear. Its not enough to have the distortion really low if the distortion content is higher orders, due to how sensitive the ear is to them. I've harped about this a lot- this is literally why tubes are still in business over 60 years after being declared 'obsolete' (that happened in the 1960s, when germanium transistors were common- on-one takes them seriously anymore- so much for 'declaration' ). Class D offers a means to amplify with benign distortion as well as tubes; so it should be no surprise they can sound smooth while being detailed at the same time (on account of lower overall distortion, which tends to obscure detail).

 

[iansr]

OK I‘m aware if the effect of a balanced topology on harmonic distortion and I should have said many class A amplifiers rather than implying ALL of them, but otherwise I think you are agreeing with the theory, assuming that by benign distortion you do indeed mean 2nd harmonic ? So I think you are saying that your class D amps have low distortion and what distortion there is has a 2nd harmonic Dominant profile. Am I right?

 

[D.Duttilleux]

OK I‘m aware if the effect of a balanced topology on harmonic distortion and I should have said many class A amplifiers rather than implying ALL of them, but otherwise I think you are agreeing with the theory, assuming that by benign distortion you do indeed mean 2nd harmonic ? So I think you are saying that your class D amps have low distortion and what distortion there is has a 2nd harmonic Dominant profile. Am I right?

You might want to read this post by Ralph (and those before and after).


In my opinion, this comment is at the core of what good audio (reproduction) entails, which is nevertheless ignored by the vast majority (especially on that other forum where SINAD is the holy grail):

"I used to be abused of the notion that there are things that you can hear that can't be measured. The problem is most of the important measurements are either never made and/or not published. Those differences you hear between various amps? Other than frequency response issues (often having to do with how the amp interfaces with the load due to output impedance), its the distortion signature that's being described (again, this is because the ear converts all forms of distortion to tonality). That is very measurable! But there is IMO a lack of awareness of what the significance of what these measurements tell us."

 

[DetroitVinylRob]

You might want to read this post by Ralph (and those before and after).


In my opinion, this comment is at the core of what good audio (reproduction) entails, which is nevertheless ignored by the vast majority (especially on that other forum where SINAD is the holy grail):

"I used to be abused of the notion that there are things that you can hear that can't be measured. The problem is most of the important measurements are either never made and/or not published. Those differences you hear between various amps? Other than frequency response issues (often having to do with how the amp interfaces with the load due to output impedance), its the distortion signature that's being described (again, this is because the ear converts all forms of distortion to tonality). That is very measurable! But there is IMO a lack of awareness of what the significance of what these measurements tell us."

Agreed.

In this idea of how and why different amplification methods acquire a sound signature and how our ears and minds process it, Ralph’s observations and possible conclusions seem to touch on a context of understanding that is simply not readily discussed nor given the recognition it is due in general audio circles.

Too often our conversations in the audio community are heavily opinionated, misinformed, and baseless. As a community we seem to have collectively forgotten much of the science and craft of the golden years of high fidelity and are confused by the modern hyperbole and focus on false identifiers of sound quality. All imho.

I am now refreshingly fascinated by these ideas and somewhat frustrated by the fact that my personal audio journey was not initially educated by this information and I have fumbled my way to ecstasy. It’s been a long road to a modest level of competency and satisfaction. But it’s good, really good to be hear.

 

[Atmasphere]

alph’s observations and possible conclusions seem to touch on a context of understanding that is simply not readily discussed nor given the recognition it is due in general audio circles.

That is because both the measurement guys and the subjectivist 'only believe your ears' guys both get really mad at the idea that there is a direct and measurable relationship between what is audible and what is measurable. It was not always like that (back in the 1980s some measurements we can do now were simply not possible or very difficult), but it is now. Add to that that quite often even if you did have the measurements available that would tell the whole story, there is a dearth of understanding of what they mean.

 

[D.Duttilleux]

That is because both the measurement guys and the subjectivist 'only believe your ears' guys both get really mad at the idea that there is a direct and measurable relationship between what is audible and what is measurable. It was not always like that (back in the 1980s some measurements we can do now were simply not possible or very difficult), but it is now. Add to that that quite often even if you did have the measurements available that would tell the whole story, there is a dearth of understanding of what they mean.

The subjectivist camp is likely too fragmented to even draw any reasonable conclusions.

In this article the subject is explored from a broader perspective.