computer audiophile on the cheap Class D amplifiers, can a chip sound as good as a regular amplifier?

[Ralf11]

still contra...

 

but tell me what to listen for in a comparison -- what does these artifacts sound like?

[GUTB]

1 minute ago, Ralf11 said:

still contra...

 

but tell me what to listen for in a comparison -- what does these artifacts sound like?

 

Different class D implementations have different sounds, but they will lack musicality / prone to listener fatigue. They can also be flat-sounding (filter destroying phase timing?). I'm talking about modern, high-end implementations. Old and cheap class D products can be downright unlistenable.

[Silly goose]

5 hours ago, GUTB said:

 

 

The answer is "no".

 

Sorry.

 

Here's the problem: switching noise. Class D amplification, currently, results in switching noise artifacts in the audio band. To defeat the switching artifacts, filtering can be used, but which ends up compromising the sound. In order to use filters that don't compromise the sound, switching has to occur at much higher frequencies than it does currently. Eventually, the technology will exist to make class D the go-to for audiophiles -- it's not here yet.

Sounds like you were frozen in a time capsule back in 2004, and was just defrosted yesterday. Which current class D amps do you have experience with? 

[GUTB]

10 minutes ago, Silly goose said:

Sounds like you were frozen in a time capsule back in 2004, and was just defrosted yesterday. Which current class D amps do you have experience with? 

 

Most recently, a Teac 301-DA, which uses ICE, and a D-Sonic M3-800S which uses a custom input stage and the latest Pascal modules. The $350 Teac was very bad, almost unlistenable. The $1,500 D-Sonic was in several ways the best amp I've ever heard, but it failed in soundstage and just couldn't get the timbre / musicality right.

[Silly goose]

Just now, GUTB said:

 

Most recently, a Teac 301-DA, which uses ICE, and a D-Sonic M3-800S which uses a custom input stage and the latest Pascal modules. The $350 Teac was very bad, almost unlistenable. The $1,500 D-Sonic was in several ways the best amp I've ever heard, but it failed in soundstage and just couldn't get the timbre / musicality right.

Well those are budget amps. The Icepower's in those amps are still using chipsets developed in 2002. Pascal's are made for sound reinforcement in pro audio applications. Have you heard anything Iceedge based yet? . How about the latest Hypex Ncore amps? The TI amps used in the Dynaudio speakers extend all the way up to 100Khz. Using a blanket statement to say all class D amps are poor, is just like saying all class A or AB is poor because you never heard good ones. 

[esldude]

5 minutes ago, GUTB said:

 

Most recently, a Teac 301-DA, which uses ICE, and a D-Sonic M3-800S which uses a custom input stage and the latest Pascal modules. The $350 Teac was very bad, almost unlistenable. The $1,500 D-Sonic was in several ways the best amp I've ever heard, but it failed in soundstage and just couldn't get the timbre / musicality right.

Do you realize the soundstage, timbre and 3D qualities of your 845 are partly or mostly an additive coloration?  Which means a high fidelity amp will have less of those.

[unbalanced output]

55 minutes ago, plissken said:

You do realize that Class a was a solution to the problem of transistor cross over distortion.

 

Lol are you trying to rewrite history? The first audio amplifier were triodes (valves if you never heard of them), which naturally operated in class A. That was invented by the Marconi company at the turn of the xxth century. Class B was invented later as a way to reduce the heat dissipation, however it sucked for audio. Quickly class AB was realised, for which there were two implementations - AB1 and AB2, the difference being essentialy on the way the valves are energised. That was all decades before the transistor was even invented - essentially it doesn't matter whether the amp is transistor or valve based, the block diagram is the similar aside from the valve supply.

[plissken]

18 minutes ago, unbalanced output said:

 

Lol are you trying to rewrite history? The first audio amplifier were triodes (valves if you never heard of them), which naturally operated in class A. That was invented by the Marconi company at the turn of the xxth century. Class B was invented later as a way to reduce the heat dissipation, however it sucked for audio. Quickly class AB was realised, for which there were two implementations - AB1 and AB2, the difference being essentialy on the way the valves are energised. That was all decades before the transistor was even invented - essentially it doesn't matter whether the amp is transistor or valve based, the block diagram is the similar aside from the valve supply.

 

That doesn't absolve it of the fact that Class A is a fully energized push/push and that is so there is no waveform distortion. That's the problem it solved in transistor amps. 

 

Class A/B and D, the other predominate audio amplification classes solve the crossover distortion PLUS efficiency, heat, and weight. 

 

Flyback and derived supplies were in use in the 30's. So what's your point again?

[esldude]

1 hour ago, GUTB said:

 

I find that hard to believe considering how massive the difference is.

This sounds familiar to other responses you have made.

 

Class D can interact negatively with some speaker loads due to the output filtering.  Then again all classes of amps need to match their performance envelop with the speaker load.  At least with many speakers there is no massive deficit in class D of good quality vs class A.   Your blanket dismissal is not convincing. 

[Jud]

2 hours ago, unbalanced output said:

Jud, I'm sorry but GUTB is right on this one. It is not a matter of "belief" that class D is an inferior approach - that's just the physical truth behind the architecture. There may be shitty class A amp implementations and good class D ones, however "inherently" as he said it is a fact fact of life that a single linear gain is superior to a quantised signal modulator. This could be easily shown mathematically. 

 

Here is some higher mathematics:

 

- Spectral Audio Class A top of the line mono amp: 98dB S/N unweighted 

 

- Mola Mola Kaluga Class D amp: 128dB S/N unweighted 

[Jud]

3 minutes ago, Jud said:

 

Here is some higher mathematics:

 

- Spectral Audio Class A top of the line mono amp: 98dB S/N unweighted 

 

- Mola Mola Kaluga Class D amp: 128dB S/N unweighted 

 

More math:

 

Pass Labs Class A Xs top of the line amp: .005% THD+N

 

Mola Mola Kaluga: .003% THD+N

[Jud]

2 hours ago, GUTB said:

 

I take this to mean that you've never listened to class A and class D amps side by side.

 

I own a Spectral Audio Class A amp.  I have heard the Mola Mola Kaluga and a custom built NCore amp at some length.

[GUTB]

6 minutes ago, Jud said:

 

I own a Spectral Audio Class A amp.  I have heard the Mola Mola Kaluga and a custom built NCore amp at some length.

 

Would you be able to easily tell the difference between the NCore and the Spectral Audio?

[Silly goose]

Here's the noise amplitude vs frequency of the TI TPA3251 class D chipset up to 50K. Where's all the switching noise in the audible band? 

 

 

[Jud]

5 minutes ago, GUTB said:

 

Would you be able to easily tell the difference between the NCore and the Spectral Audio?

 

I'm tempted after that experience to build an nCore amp and sell the Spectral.

[esldude]

22 minutes ago, Silly goose said:

Here's the noise amplitude vs frequency of the TI TPA3251 class D chipset up to 50K. Where's all the switching noise in the audible band? 

 

 

The kind of debilitating noise referred to here is special audiophile noise that doesn't show up in measurements.  Class D amps only put out this noise when audiophiles are listening without the amp or the listener being tested.  Otherwise it disappears. 

[esldude]

19 minutes ago, Jud said:

 

I'm tempted after that experience to build an nCore amp and sell the Spectral.

So you might join the Dark side?  Why do you think they call it class D?

[GUTB]

 

[Silly goose]

Here's the Icepower amps that use 15 year old technology most audiophiles think are the pinnacle of class D performance today. Look at the massive noise rising rapidly at the end of the audible band, and phase shift into the audible band. Then compare with the modern TI chip:

 

[Ralf11]

14 minutes ago, esldude said:

The kind of debilitating noise referred to here is special audiophile noise that doesn't show up in measurements.  Class D amps only put out this noise when audiophiles are listening without the amp or the listener being tested.  Otherwise it disappears. 

 

 

it ain't NOISE

 

it's them arty-facts

[esldude]

34 minutes ago, Silly goose said:

Here's the Icepower amps that use 15 year old technology most audiophiles think are the pinnacle of class D performance today. Look at the massive noise rising rapidly at the end of the audible band, and phase shift into the audible band. Then compare with the modern TI chip:

 

I don't know about that.  Yes some ICEpower amps were like that.  Like any class of operation it depends upon the details.  Or are you claiming there are no improvements in the ICEpower modules in the last 15 years. 

 

You can see Bel Canto Icepower amps from 2008 tested here:

http://www.stereophile.com/content/bel-canto-eone-s300iu-integrated-amplifier-measurements#rbssE3SW58uxZyEI.97

 

Still have their problems, but an improvement.  Also an improvement over effects of loading compared to even earlier Tripath amps from TI.  Your left hand graph above looks like an early Tripath actually instead of Icepower, but I don't know the source of your graph. Some Icepower amps did look like that.  Others looked better than the Bel Canto. And two different graphs the one on the right doesn't really tell what the TI chip does compared to the different graph on the left.

 

There have been several revisions since then and makers have improved how they filter the output.  Plus you now have Ncore Hypex to choose from if you wish to use class D. Of course the key improvement by Bruno Putzeys is to include the load as part of the filter for the output.  This lets the load not effect the response and phase of the speaker so much.  So you get something like this.

 

[Silly goose]

22 minutes ago, esldude said:

I don't know about that.  Yes some ICEpower amps were like that.  Like any class of operation it depends upon the details.  Or are you claiming there are no improvements in the ICEpower modules in the last 15 years. 

 

You can see Bel Canto Icepower amps from 2008 tested here:

http://www.stereophile.com/content/bel-canto-eone-s300iu-integrated-amplifier-measurements#rbssE3SW58uxZyEI.97

 

Still have their problems, but an improvement.  Also an improvement over effects of loading compared to even earlier Tripath amps from TI.  Your left hand graph above looks like an early Tripath actually instead of Icepower, but I don't know the source of your graph. Some Icepower amps did look like that.  Others looked better than the Bel Canto. And two different graphs the one on the right doesn't really tell what the TI chip does compared to the different graph on the left.

 

There have been several revisions since then and makers have improved how they filter the output.  Plus you now have Ncore Hypex to choose from if you wish to use class D. Of course the key improvement by Bruno Putzeys is to include the load as part of the filter for the output.  This lets the load not effect the response and phase of the speaker so much.  So you get something like this.

 

I'm saying that the class D the class d naysayers are talking about is what I've shown there. And that's their latest 700WPC amp the ASC700. Until the Iceedge chipset based amps are released, the full lineup is still using the circa 2002 ICCX chipset. 

 

I've heard several of the latest class D amps. Hypex NC400,NC500, NC1200. 6 different Icepower. Anaview, Pascal. The Ncores are very good, but this little TI 3251 chip really impressed me. Especially @ $4.44 a piece in lots of 1000 units. As far as I'm concerned, anything but class D is only fit for boat anchors these days. And they are only going to keep improving. 

 

 

 

[Silly goose]

Sweatshop supporters will likely be impressed with this unit:

 

https://www.aliexpress.com/store/product/G3H-300W-TPA3251PurePath-Ultra-HD-Analog-Input-Class-D-Amplifier-integrated-with-global-AC-supply/601158_32712158006.html

[esldude]

43 minutes ago, Silly goose said:

Sweatshop supporters will likely be impressed with this unit:

 

https://www.aliexpress.com/store/product/G3H-300W-TPA3251PurePath-Ultra-HD-Analog-Input-Class-D-Amplifier-integrated-with-global-AC-supply/601158_32712158006.html

 

Everyone who reads CA and lurks must have ordered one.  Your link and others all show it not available now.  Could be that Mark Levinson and Jeff Rowland purchased them all to rebox into their line up. 

[Silly goose]

Just now, esldude said:

 

Everyone who reads CA and lurks must have ordered one.  Your link and others all show it not available now.  Could be that Mark Levinson and Jeff Rowland purchased them all to rebox into their line up. 

Likely. Why go through all the work to make all the internals when folks are only going to buy it for the case anyways.