Class D Amps, the Ins & Outs

[Allan]

I've been thinking of buying a Class D amp for summer as my heavy weight Class A and A/ABs get quite warm and can heat the room.

 

I'd like to know the ins and outs of these things. There's a few different brand Class D modules, like the Hypex NCx500 or the Purifi 1ET7040 and 1ET400A modules. Is there any difference between these, does one particlular type have benefits over the other?!?

 

What are the sonic characters of Class D in general?!?

 

I've ProAc and Harbeth speakers, is there a particular type and or brand of Class D amp that would be better suited to these speakers?!?

 

Also quite a few Class D amps I've seen use SMPS (switch mode power supplies) , is there a brand available that uses Transformers with CRC.. or CLC..??

Has any compared a SMPS Class D design to a Transformer design, was there any sonic benefit or difference?!?

 

Thanks AL

 

 

 

Edited November 3, 2024 by Allan

[Steff]

Here's an informative write-up on the FOSI amps: https://archimago.blogspot.com/2024/08/part-ii-fosi-audio-v3-mono-amp-class-d.html

[Ahh- Schnoo Schnoo]

I saved this from a review, as it described what I heard.

This is from some time ago. Haven't listened to the latest Class D iterations.

Cheers

 

Class-D creates a simulacrum of the original signal, which feels slightly spartan and texturally different than class-A or -AB. Class-D generates power easily into a wide range of speaker loads. It can be superdynamic. It appears cleaner and more transparent than class-A. Higher in contrast. Quiet passages and empty spaces can feel a bit electronic. Images are well-outlined but devoid of warm flesh. Voices and instruments display a saturated but distinctly limited palette of tones. The feedback and low-pass filtering required for class-D add hardness to the sound. The best-executed class-D amplifiers deliver music with a refined, grainless quality that many audiophiles find appealing.

Class-A is naturally low distortion. Maximum information-density. Less distinctly drawn and impactful than class-D. More overtly textured, painterly, colorful, and full-bodied. Lower in contrast than class-D but with a wider, more complex palette of tones. Empty spaces seem charged and appear connected to performers. The lower feedback required to achieve linearity allows for a more supple and relaxed presentation. Properly executed class-A sounds trippy and lifelike.

Class-AB is a compromise that allows amplifiers to produce class-A power up to some predetermined bias point (typically no more than a few watts), at which point it crosses over to almost-class-B, thereby allowing more watts (of a reduced quality) before apocalyptic clipping sets in. Driving benign loads of reasonable sensitivity, class-AB amplifiers, especially some tube models, can and sometimes will sound superbly class-A-like. Hence their popularity.

[Art Vandelay]

I recently finished assembly of NCx500 monoblocks. They can sound a bit brash but I've found that any slight harshness is eliminated with a ferrite clamp-on  at the audio input connector.  I'm driving with input buffers disabled, so just the 11.5dB of gain. 

 

Overall, very detailed and transparent, and better than many class AB amplifiers I've heard, including some expensive ones, and they run pretty much stone cold, so ideal for summer months.  

 

 

[Snoopy8]

Maybe this can help?  By the Class D master himself...

 

 

 

[aussievintage]

5 hours ago, Allan said:

Also quite a few Class D amps I've seen use SMPS (switch mode power supplies) , is there a brand available that uses Transformers with CRC.. or CLC..??

Has any compared a SMPS Class D design to a Transformer design, was there any sonic benefit or difference?!?

 

The best in class use SMPS.  

[Allan]

50 minutes ago, aussievintage said:

 

The best in class use SMPS.  

Well I would have to agree a SMPS can perform very well.

 

I have a Pass ACA that uses one, it performs remarkably well compared to Krell and Mark Levinson amps I have.. so much so its my prefered amp to use with my Harbeths SHL5 speakers.. obviously if listening levels don't demand big numbers.

 

It just doent make sense as the Harbeths are reputed to love power and gobs of current, yet the ACA is so sweet sounding its ridiculously addictive. It's only 5W of pure Class A whereas the Krell is 150W high bias A/AB and the Levinson is 100W AB

 

Anyone who has the Harbeths really owes themselves the opportunity to try the ACA on them. They can be snapped up new or used for under $500!

 

So if I can find a Class D amp around 200w in 8ohm that has sonic performance somewhere between the ACA and the Krell and ML, I'm in

 

 

Edited November 3, 2024 by Allan

[Irek]

3 hours ago, Allan said:

Well I would have to agree a SMPS can perform very well.

 

I have a Pass ACA that uses one, it performs remarkably well compared to Krell and Mark Levinson amps I have.. so much so its my prefered amp to use with my Harbeths SHL5 speakers.. obviously if listening levels don't demand big numbers.

 

It just doent make sense as the Harbeths are reputed to love power and gobs of current, yet the ACA is so sweet sounding its ridiculously addictive. It's only 5W of pure Class A whereas the Krell is 150W high bias A/AB and the Levinson is 100W AB

 

Anyone who has the Harbeths really owes themselves the opportunity to try the ACA on them. They can be snapped up new or used for under $500!

 

So if I can find a Class D amp around 200w in 8ohm that has sonic performance somewhere between the ACA and the Krell and ML, I'm in

 

 

I have 2 amps, Primare i25 class D and Primaluna Evo100. Both are excellent but also very different. 

[LogicprObe]

 

It's always a dilema until you listen..........no matter what the class.

[David A]

8 hours ago, aussievintage said:

 

The best in class use SMPS.  

 

I've found the best sounding class D amps tend to use toroidal transformers with GaNFET output devices. Though YMMV.
 

Edited November 3, 2024 by David A

[Allan]

7 hours ago, David A said:

 

I've found the best sounding class D amps tend to use toroidal transformers with GaNFET output devices. Though YMMV.
 

a very interesting comment Dave and one that I'd hoped would be raised without me leading to it. What attribute was it that took you to the conclusion the toroidal powered amps sounded best please?

 

The GaNFet's seem to be getting great reports. AGD has some sexy GaFet amps.

 

I'm also considering building the Class D if I can't find a commercial version of what I want. My preferences in design are Class A input buffer and the possibility of a toroidal supply. GaFet now also getting some serious consideration.. thanks @David A

 

A GaNet writeup, well worth a read

 

I see Mytek have a toroidal powered GaFet Class D amp out

 

 

 

 

Edited November 3, 2024 by Allan

[Snoopy8]

3 hours ago, Allan said:

a very interesting comment Dave and one that I'd hoped would be raised without me leading to it. What attribute was it that took you to the conclusion the toroidal powered amps sounded best please?

 

The GaNFet's seem to be getting great reports. AGD has some sexy GaFet amps.

 

I'm also considering building the Class D if I can't find a commercial version of what I want. My preferences in design are Class A input buffer and the possibility of a toroidal supply. GaFet now also getting some serious consideration.. thanks @David A

 

A GaNet writeup, well worth a read

 

I see Mytek have a toroidal powered GaFet Class D amp out

 

Have you read this? 

https://icepoweraudio.com/is-gan-worth-it/

 

Also in Bruno Putzey's article (that I posted above) in the section called "The Revolution that is, and isn’t GaN".

 

[sir sanders zingmore]

A bit sad to see the old misconceptions about class D still being trotted out in some of the linked articles. 
It’s come a long way. 

[Snoopy8]

1 hour ago, sir sanders zingmore said:

A bit sad to see the old misconceptions about class D still being trotted out in some of the linked articles. 
It’s come a long way. 

Until more people read and understand the Bruno Putzeys article, the misconceptions here on Class D will unfortunately continue... 

[aussievintage]

7 hours ago, Allan said:

a very interesting comment Dave and one that I'd hoped would be raised without me leading to it.

 

Beware of seeking advice that confirms your own already held preferences.    Class D technology is undergoing rapid development, and that also often leaves companies (and people) defending outdated ideas.

[David A]

9 hours ago, Allan said:

a very interesting comment Dave and one that I'd hoped would be raised without me leading to it. What attribute was it that took you to the conclusion the toroidal powered amps sounded best please?

 

The GaNFet's seem to be getting great reports. AGD has some sexy GaFet amps.

 

I'm also considering building the Class D if I can't find a commercial version of what I want. My preferences in design are Class A input buffer and the possibility of a toroidal supply. GaFet now also getting some serious consideration.. thanks @David A

 

I'm not a tech guy or DIY'er. So my comment is/was based on my research into class D amps which, for my tastes, sound more tube-like and musical vs the stereotypical dry flat class D sound. And the best examples I found are AGD and Atma-Sphere. Though it should be noted that Ralph Karsten's Class D mono's use a circuit entirely of his own design and manufacture, and do not use anyone else's modules. So Ralph was able to tune the circuit to sound like his OTL valve amps. Indeed they're the amps he uses in his own system.

 

[David A]

2 hours ago, aussievintage said:

 

Beware of seeking advice that confirms your own already held preferences.    Class D technology is undergoing rapid development, and that also often leaves companies (and people) defending outdated ideas.

 

Yes confirmation bias is a thing.

[Allan]

Yes AGD amps sure have my attention, if they sound half as good as they look....

[Art Vandelay]

16 hours ago, Allan said:

 

 

I'm also considering building the Class D if I can't find a commercial version of what I want. My preferences in design are Class A input buffer and the possibility of a toroidal supply.

 

 

 

 

 

 

Best I've heard so far employ a real class A buffer. Op-amp based is fine, but if you want real class A you need to add a transistor or mosfet buffer stage with >10mA quiescent current.  Op-amps run out of class A beyond 1-2mA. 

[Allan]

20 minutes ago, Art Vandelay said:

 if you want real class A you need to add a transistor or mosfet buffer stage....

 

and why not Class A buffer, its cheap

 

 

Edited November 4, 2024 by Allan

[Art Vandelay]

4 minutes ago, Allan said:

 

and why not Class A buffer, its cheap

 

 

 

 

A few more components and a small amount of heat. I'm currently driving my NCx500's direct from the dac but I'll get around to building a dedicated class A buffer with selectable gain. Probably build into a separate enclosure with a linear power supply so that it's isolated from the high frequency switching noise. 

 

 

[Allan]

34 minutes ago, Art Vandelay said:

Probably build into a separate enclosure with a linear power supply so that it's isolated from the high frequency switching noise. 

 

 

I'm hearing you. Building a Pearl 3 phono ATM.. same theory. 1st watt integrity, quality from there 

 

 

Edited November 4, 2024 by Allan

[aussievintage]

It's funny how we go back to the old ideas and design.  I mean,   nearly all preamp/driver/buffers in the valve era were all class A.  Early transistor small signal circuits were class A as well.   It's just the easiest simplest circuit.

 

You guys should put a nice clean triode in there for a buffer.

[Allan]

5 hours ago, aussievintage said:

It's funny how we go back to the old ideas and design.  I mean,   nearly all preamp/driver/buffers in the valve era were all class A.  Early transistor small signal circuits were class A as well.   It's just the easiest simplest circuit.

I think that can be said about a lot of tech. Its weird how the original means to present a concept carries through future development... it's fair to say they nailed it right off the bat

[Art Vandelay]

9 hours ago, aussievintage said:

It's funny how we go back to the old ideas and design.  I mean,   nearly all preamp/driver/buffers in the valve era were all class A.  Early transistor small signal circuits were class A as well.   It's just the easiest simplest circuit.

 

You guys should put a nice clean triode in there for a buffer.

 

 Hypex or Purifi modules require 10-15vrms and present an input Z of 1.8k ohms, so it's not as easy as driving a normal power amplifier with typically >20k input impedance at 2vrms. Most op-amp based preamplifiers will drive a (typical) power amplifier without excursion from class A region, so it's generally not necessary to include an additional  class A buffer.       

If you want to build a tube based buffer for Hypex / Purifi you'll probably need to add a solid state output stage with DC servo etc, because the Hypex module's performance will be compromised if the input isn't balanced and referenced to earth, which rules out transformer and AC coupling, so that's going to be a show stopper for many tube lovers.