Curious about Class D Amplifiers: Hypex Ncore, B&O ICE, etc.

[chicagorspec]

Feb 6, 2018 13:01:07 GMT -5 mgbpuff said:

Walking the Class D doggie. Note the output filter (see mouth) required for safe playing with this kind of doggie architecture.

You’re just so witty and relevant.

[pedrocols]

Just noticed a different theme on the board looks.

[mgbpuff]

Feb 6, 2018 13:16:01 GMT -5 chicagorspec said:

Feb 6, 2018 13:01:07 GMT -5 mgbpuff said:

Walking the Class D doggie. Note the output filter (see mouth) required for safe playing with this kind of doggie architecture.

You’re just so witty and relevant.

Yeh, yeh, I know - vs a real doggie, it doesn't put out much heat either; it's cold blooded. And its mouth is either open or closed, nothing in between.

[klinemj]

Feb 6, 2018 13:01:07 GMT -5 mgbpuff said:

Walking the Class D doggie. Note the output filter (see mouth) required for safe playing with this kind of doggie architecture.

So, you don't care for its output filter, so you put the whole class of amps down? Well...alrighty then...you don't like then, don't buy them. But I see no reason to put them down when someone (Bruno P) has figured it out. Personally, I don't care what it takes to make any piece of gear sound good and work well if its cost and performance meet my needs. And, the Hypex nCores more than meet my needs. Not to disrespect Emotiva at all with the statement to come, but I did get rid of my XPA-2 because I felt the Hypex sounded better.

Mark

[mgbpuff]

Feb 8, 2018 8:23:47 GMT -5 klinemj said:

Feb 6, 2018 13:01:07 GMT -5 mgbpuff said:

Walking the Class D doggie. Note the output filter (see mouth) required for safe playing with this kind of doggie architecture.

So, you don't care for its output filter, so you put the whole class of amps down? Well...alrighty then...you don't like then, don't buy them. But I see no reason to put them down when someone (Bruno P) has figured it out. Personally, I don't care what it takes to make any piece of gear sound good and work well if its cost and performance meet my needs. And, the Hypex nCores more than meet my needs. Not to disrespect Emotiva at all with the statement to come, but I did get rid of my XPA-2 because I felt the Hypex sounded better.

Mark

I don't care for the whole approach of Class D design; it is a Dr. Frankenstein approach to approximating a human. For equivalent efficiency, I would like to see a true Class H perfected (a tracking power supply [ a la Bob Carver], not just a two step power supply), because Class A/B is really what an amplifier is all about - something that simply amplifies an analog incoming signal, not something that dissects, reconfigures, and then reassembles. All that crap is not necessary at all, and causes all sorts of other problems including radiations, noise in other circuits, and sensitivity to different loads. To make Class D work is like another Manhattan Project just to make a better (strike that for different) automobile. To me, it represents brilliant engineering that is grossly miss directed and wasted for no real gain.

[405x5]

Class D perfect for a subwoofer plate amplifier.....saves the day for cost effective service.

Solid state A/B handles the rest....I’m delighted 😃

Bill

[mgbpuff]

Only because sympathetic resonance in most house structures (unless you live in an underground concrete bomb shelter) disguises low frequency audio performance and deems accuracy moot.

[KeithL]

I hate to break it to you, but the closest and most "direct" form of sound recording and reproduction to how we humans hear is a mechanical cylinder recording.
The sound pressure in the air pushes on a diaphragm, which pushes a needle, which cuts the plastic; then, when you play it back, the plastic pushes a needle, which pushes a diaphragm, which pushes the air.
It's all "dead simple", and there is no conversion between mechanical displacement and an electrical signal at all.
Unfortunately nobody ever managed to get the system to produce very good results.... but it was "simple and very direct".

I'm inclined to agree with you - that Class A/B does the job very well.
However, Class D amplifiers are a lot more efficient, which means both less power consumed, and less heat to get rid of... and so usually smaller and lighter boxes.
(Some people consider this important.)
In the past, Class D amps tended to not sound very good... but many recent designs sound just fine.

Unfortunately, your idea about "a full tracking power supply" is sort of a red herring......
Most Class H amplifiers switch between multiple rails (so they do switch).
And, if, instead of switching, you want to have the power supply rails track the signal in a "non-switching sort of way" then you're simply back to Class A/B (more like something called "cascode" - which is about as efficient as Class A/B).

Using a "variable switching supply" to "move the rails to follow the signal" is a pretty good description of what a Class D amplifier does to begin with.
So what you're really talking about is "using a Class D amplifier to run a Class A/B amplifier" - which is actually more complicated than either one is separately.
That idea made good sense back when Class D amplifiers were very efficient - but didn't sound very good.
However, today, it's sort of the long way around....
If you're going to include the complexity and parts count of a Class D amplifier in your design anyway, you don't gain much by running a Class A/B linear output stage off of its output.
(It's about equivalent to loading your car onto a flat bed train car to drive cross country "because the train gets better mileage but the car is more comfortable".)

(There was also a time when you could argue that adding computers to cars wasn't really worth the extra complexity - because carburetors worked just fine - but that just isn't true today.)

Feb 8, 2018 9:24:04 GMT -5 mgbpuff said:

Feb 8, 2018 8:23:47 GMT -5 klinemj said:

So, you don't care for its output filter, so you put the whole class of amps down? Well...alrighty then...you don't like then, don't buy them. But I see no reason to put them down when someone (Bruno P) has figured it out. Personally, I don't care what it takes to make any piece of gear sound good and work well if its cost and performance meet my needs. And, the Hypex nCores more than meet my needs. Not to disrespect Emotiva at all with the statement to come, but I did get rid of my XPA-2 because I felt the Hypex sounded better.

Mark

I don't care for the whole approach of Class D design; it is a Dr. Frankenstein approach to approximating a human. For equivalent efficiency, I would like to see a true Class H perfected (a tracking power supply [ a la Bob Carver], not just a two step power supply), because Class A/B is really what an amplifier is all about - something that simply amplifies an analog incoming signal, not something that dissects, reconfigures, and then reassembles. All that crap is not necessary at all, and causes all sorts of other problems including radiations, noise in other circuits, and sensitivity to different loads. To make Class D work is like another Manhattan Project just to make a better (strike that for different) automobile. To me, it represents brilliant engineering that is grossly miss directed and wasted for no real gain.

[annjones13]

So is the rumored Emersa class D amp still under consideration at Emotiva?
Ann

[mgbpuff]

Feb 8, 2018 10:55:56 GMT -5 KeithL said:

I hate to break it to you, but the closest and most "direct" form of sound recording and reproduction to how we humans hear is a mechanical cylinder recording.
The sound pressure in the air pushes on a diaphragm, which pushes a needle, which cuts the plastic; then, when you play it back, the plastic pushes a needle, which pushes a diaphragm, which pushes the air.
It's all "dead simple", and there is no conversion between mechanical displacement and an electrical signal at all.
Unfortunately nobody ever managed to get the system to produce very good results.... but it was "simple and very direct".

I'm inclined to agree with you - that Class A/B does the job very well.
However, Class D amplifiers are a lot more efficient, which means both less power consumed, and less heat to get rid of... and so usually smaller and lighter boxes.
(Some people consider this important.)
In the past, Class D amps tended to not sound very good... but many recent designs sound just fine.

Unfortunately, your idea about "a full tracking power supply" is sort of a red herring......
Most Class H amplifiers switch between multiple rails (so they do switch).
And, if, instead of switching, you want to have the power supply rails track the signal in a "non-switching sort of way" then you're simply back to Class A/B (more like something called "cascode" - which is about as efficient as Class A/B).

Using a "variable switching supply" to "move the rails to follow the signal" is a pretty good description of what a Class D amplifier does to begin with.
So what you're really talking about is "using a Class D amplifier to run a Class A/B amplifier" - which is actually more complicated than either one is separately.
That idea made good sense back when Class D amplifiers were very efficient - but didn't sound very good.
However, today, it's sort of the long way around....
If you're going to include the complexity and parts count of a Class D amplifier in your design anyway, you don't gain much by running a Class A/B linear output stage off of its output.
(It's about equivalent to loading your car onto a flat bed train car to drive cross country "because the train gets better mileage but the car is more comfortable".)

(There was also a time when you could argue that adding computers to cars wasn't really worth the extra complexity - because carburetors worked just fine - but that just isn't true today.)

Feb 8, 2018 9:24:04 GMT -5 mgbpuff said:

I don't care for the whole approach of Class D design; it is a Dr. Frankenstein approach to approximating a human. For equivalent efficiency, I would like to see a true Class H perfected (a tracking power supply [ a la Bob Carver], not just a two step power supply), because Class A/B is really what an amplifier is all about - something that simply amplifies an analog incoming signal, not something that dissects, reconfigures, and then reassembles. All that crap is not necessary at all, and causes all sorts of other problems including radiations, noise in other circuits, and sensitivity to different loads. To make Class D work is like another Manhattan Project just to make a better (strike that for different) automobile. To me, it represents brilliant engineering that is grossly miss directed and wasted for no real gain.

Over complexity is of course undesirable, but the ideal would be to be able to predetermine the level that would be needed for an oncoming signal. It wouldn't have to be an exact image of the signal, just a level that would be out of the way for a predetermined period. Perhaps an embedded or parallel signal that represents the power level that will shortly be required (that way video coordinated sound would not be deteriorated For just sound reproduction, the music signal could just be delayed so the power level to be needed is adjusted first. Or the delay could be turned on or off depending on whether coordination with video is required.

[mgbpuff]

On second thought, maybe Class H with a switching power supply is good enough. Hey, isn't that what Emotiva has now?

[KeithL]

Such ideas have been around for a while...

Since you can't see the future, the only way to do that would be to delay the audio signal, so the amplifier could "look ahead" and adjust its power supply rails in advance.

This means that:
1) You would have to introduce a delay in all audio signals your amplifier is going to play.
1a) And, yes, if there was video involved, you would have to delay that as well.
2) You're going to have to convert everything it's going to play into a digital audio signal (there's no such thing as an analog delay that doesn't have horrible side effects).
2a) Alternately, you could create and include a separate "control track", which would have to be stored with, and played in synch with, the actual audio track.
2b) (Of course, this means that your wonder-amp would only be able to play audio that had been "pre-processed" specifically for it.)

It all sounds rather........ intricate and complicated.
In fact, Class D amplifiers look simple by comparison.
(A basic Class D amplifier is actually pretty simple.... the complexity is in the details to get it to work well.)

I should also note that, in a digital audio system, where you're playing a file that's stored locally, "looking ahead" is pretty trivial (audio editor software does it routinely).
Therefore, if you were combining your player (which has access to the source file), and your amplifier, this would be pretty easy to do.
However, it's still a lot EASIER to just use a Class D amplifier, which is going to deliver efficiency on par with what this system would do at its best anyway.
(And that's exactly what most people do....  and it's a lot easier to combine a good file player with a good separate amplifier than it is to tie them both together 8 ways from Sunday.)

There's also one other little detail that everyone seems to overlook......
The performance of a Class A/B amplifier VARIES DEPENDING ON THE RAIL VOLTAGE.
You can't just "move the rails up and down independent of everything else" - because the electrical characteristics of the individual transistors vary depending on the voltage across them.
And you have to compensate for those changes elsewhere in the design.
(Which is why even "simple Class H" isn't quite as simple as you might think.)

In short, while these ideas would all make great PhD papers, most of them are way too complicated to bother with in a commercial product.

<abbr class="o-timestamp time" data-timestamp="1518108202000" title="Feb 8, 2018 10:43:22 GMT -6">Feb 8, 2018 10:43:22 GMT -6</abbr> mgbpuff said:

Feb 8, 2018 11:43:22 GMT -5 mgbpuff said:

Feb 8, 2018 10:55:56 GMT -5 KeithL said:

I hate to break it to you, but the closest and most "direct" form of sound recording and reproduction to how we humans hear is a mechanical cylinder recording.
The sound pressure in the air pushes on a diaphragm, which pushes a needle, which cuts the plastic; then, when you play it back, the plastic pushes a needle, which pushes a diaphragm, which pushes the air.
It's all "dead simple", and there is no conversion between mechanical displacement and an electrical signal at all.
Unfortunately nobody ever managed to get the system to produce very good results.... but it was "simple and very direct".

I'm inclined to agree with you - that Class A/B does the job very well.
However, Class D amplifiers are a lot more efficient, which means both less power consumed, and less heat to get rid of... and so usually smaller and lighter boxes.
(Some people consider this important.)
In the past, Class D amps tended to not sound very good... but many recent designs sound just fine.

Unfortunately, your idea about "a full tracking power supply" is sort of a red herring......
Most Class H amplifiers switch between multiple rails (so they do switch).
And, if, instead of switching, you want to have the power supply rails track the signal in a "non-switching sort of way" then you're simply back to Class A/B (more like something called "cascode" - which is about as efficient as Class A/B).

Using a "variable switching supply" to "move the rails to follow the signal" is a pretty good description of what a Class D amplifier does to begin with.
So what you're really talking about is "using a Class D amplifier to run a Class A/B amplifier" - which is actually more complicated than either one is separately.
That idea made good sense back when Class D amplifiers were very efficient - but didn't sound very good.
However, today, it's sort of the long way around....
If you're going to include the complexity and parts count of a Class D amplifier in your design anyway, you don't gain much by running a Class A/B linear output stage off of its output.
(It's about equivalent to loading your car onto a flat bed train car to drive cross country "because the train gets better mileage but the car is more comfortable".)

(There was also a time when you could argue that adding computers to cars wasn't really worth the extra complexity - because carburetors worked just fine - but that just isn't true today.)

Over complexity is of course undesirable, but the ideal would be to be able to predetermine the level that would be needed for an oncoming signal. It wouldn't have to be an exact image of the signal, just a level that would be out of the way for a predetermined period. Perhaps an embedded or parallel signal that represents the power level that will shortly be required (that way video coordinated sound would not be deteriorated For just sound reproduction, the music signal could just be delayed so the power level to be needed is adjusted first. Or the delay could be turned on or off depending on whether coordination with video is required.

[KeithL]

By Jove, I think you've got it.....

Feb 8, 2018 12:04:19 GMT -5 mgbpuff said:

On second thought, maybe Class H with a switching power supply is good enough. Hey, isn't that what Emotiva has now?

[klinemj]

Feb 8, 2018 9:24:04 GMT -5 mgbpuff said:

Feb 8, 2018 8:23:47 GMT -5 klinemj said:

So, you don't care for its output filter, so you put the whole class of amps down? Well...alrighty then...you don't like then, don't buy them. But I see no reason to put them down when someone (Bruno P) has figured it out. Personally, I don't care what it takes to make any piece of gear sound good and work well if its cost and performance meet my needs. And, the Hypex nCores more than meet my needs. Not to disrespect Emotiva at all with the statement to come, but I did get rid of my XPA-2 because I felt the Hypex sounded better.

Mark

I don't care for the whole approach of Class D design; it is a Dr. Frankenstein approach to approximating a human. For equivalent efficiency, I would like to see a true Class H perfected (a tracking power supply [ a la Bob Carver], not just a two step power supply), because Class A/B is really what an amplifier is all about - something that simply amplifies an analog incoming signal, not something that dissects, reconfigures, and then reassembles. All that crap is not necessary at all, and causes all sorts of other problems including radiations, noise in other circuits, and sensitivity to different loads. To make Class D work is like another Manhattan Project just to make a better (strike that for different) automobile. To me, it represents brilliant engineering that is grossly miss directed and wasted for no real gain.

If it sounds great, who cares (especially when it is very light, compact, low energy consumption, runs cool, and is cost effective)? As for whether it was a Manhattan Project...the best results of Class D have been by the work of one man over a fairly short period of time. Hardly equivalent to a Manhattan Project. And, from all the extremely positive reviews of the result of his efforts - it's a real stretch to say there was "no real gain".

But, I never expect to hear you admit that - which is fine.

Mark

[Casey Leedom]

By the way, speaking on Ncore-based amplifiers, has anyone looked at the ATI AT527NC? It looks like one could get a 7-Channel unit for $4k ...

Casey