Are there Class D amps that eliminate the original treble complaints?

[davide256]

Wondering what the current state of Class D amps is? A year ago I replaced my aging  class AB amp with Orchard Audio GaN mono blocks, which

was a major improvement in bass foundation and low volume resolution. After living with these for a year still pretty pleased but have realized that in high treble

they could be improved on, as they can accentuate irritants causing me to keep a tube pre to smooth them out. Reading reviews in the past, this seemed to be the

one criticism of class D. I'm wondering if any newer technology has emerged for class D that resolves this?

[Apollo]

When I was looking for a class D amp 2 years ago, I first tried the Mola Mola Kaluga monoblocks. They unfortunately had real issues in the higher frequencies, sounding very harsh at times.

I then tried (and bought) Apollon Audio monoblocks, using the same Hypex PNC1200 amp modules, but proceeded with a Weiss  Op2-bp opamp input buffer.

All high frequency problems gone. 
They have replaced Ayre MX-R Twenty monoblocks (very sweet).

 

Now you can have the same amplifiers using Purify highpower modules, which should better the Hypex modules especially in the higher frequencies.

I have no links whatsoever with Apollon audio. I chose my username here, just because Inlike my amps that much.

 

[ssh]

• @barrows has praised  the Apollon amps several times on this forum. I replaced my Yiggy/Raggy combo with a Bricasti M3 directly connected to a Apollon 1ET400A Mini Stereo Amplifier.  Quite satisfied with my change. Now I need to get me some Salk BePure speakers. The six month wait is a killer though.

[barrows]

1 hour ago, Apollo said:

When I was looking for a class D amp 2 years ago, I first tried the Mola Mola Kaluga monoblocks. They unfortunately had real issues in the higher frequencies, sounding very harsh at times.

I then tried (and bought) Apollon Audio monoblocks, using the same Hypex PNC1200 amp modules, but proceeded with a Weiss  Op2-bp opamp input buffer.

All high frequency problems gone. 
They have replaced Ayre MX-R Twenty monoblocks (very sweet).

 

Now you can have the same amplifiers using Purify highpower modules, which should better the Hypex modules especially in the higher frequencies.

I have no links whatsoever with Apollon audio. I chose my username here, just because Inlike my amps that much.

 

That's an odd point of view...  Every time I have heard the Kalugas they sounded fantastic, perhaps there was a system matching problem.  Anyway, I am glad you found a solution which worked for you.  Generally speaking, I find the Purifi modules improve over NCore based solutions mostly in the high frequencies, and the difference is not huge, with the Purifi modules having a bit more transparency and perhaps air in the top bands.

[Cormorant]

I've had a pair of the new Hypex Nilai 500 monoblocks in my system since the beginning of January.

 

I find the treble to be superb -- extended, very clear, and tonally excellent.  The smoothness really makes one appreciate low distortion in the treble range.

 

I have not heard any Purifi amps in my system to compare.

[davide256]

2 hours ago, Cormorant said:

I've had a pair of the new Hypex Nilai 500 monoblocks in my system since the beginning of January.

 

I find the treble to be superb -- extended, very clear, and tonally excellent.  The smoothness really makes one appreciate low distortion in the treble range.

 

I have not heard any Purifi amps in my system to compare.

hmm, ASR likes them. But of course they only measure without opining on sound quality :<}

 

https://www.audio “science” review/forum/index.php?threads/hypex-nilai500diy-amplifier-review.41669/

 

[Miska]

9 hours ago, davide256 said:

hmm, ASR likes them. But of course they only measure without opining on sound quality :<}

 

I would try something with Purifi modules. Like for example recent more expensive NAD models. These don't have as much output impedance issue at high frequencies, so they are less sensitive to speaker pairing. Class-D amps are a bit like tube power amps, where the speaker pairing plays bigger role.

 

Two things however to take into account. First the Hypex/Purifi models are not particularly great for low impedance difficult loads. So if a speaker impedance dips below 4 ohm or has low EPDR impedance, those amps can collapse. Second thing to pay attention to is the kind of DAC / filters paired with the DAC. So I would avoid running leaky filters or R2R at low NOS rates, because class-D amplifiers can alias the high level images down to audio band.

 

[davide256]

45 minutes ago, Miska said:

 

I would try something with Purifi modules. Like for example recent more expensive NAD models. These don't have as much output impedance issue at high frequencies, so they are less sensitive to speaker pairing. Class-D amps are a bit like tube power amps, where the speaker pairing plays bigger role.

 

Two things however to take into account. First the Hypex/Purifi models are not particularly great for low impedance difficult loads. So if a speaker impedance dips below 4 ohm or has low EPDR impedance, those amps can collapse. Second thing to pay attention to is the kind of DAC / filters paired with the DAC. So I would avoid running leaky filters or R2R at low NOS rates, because class-D amplifiers can alias the high level images down to audio band.

 

 Magnepan's do dip to 2 ohms above 6khz.  The treble issues with current class D amp are actually at their worst playing LP's vs my digital solution{s} with same music.

Would hate to give up the resolution of class D as its so much better than my prior tube and class AB amps from the lower treble down,

but somethings  got to change so I don't cringe at Enya, choral music and "Close Encounters of the Third Kind"

[Miska]

26 minutes ago, davide256 said:

 Magnepan's do dip to 2 ohms above 6khz.  The treble issues with current class D amp are actually at their worst playing LP's vs my digital solution{s} with same music.

 

LP will spit out lot of harmonics beyond 20 kHz, because the cartridge THD near 20 kHz is pretty high. It already begins to increase from about 6 kHz. Probably not high enough to cause aliasing in the amp, but enough to cause increased power draw at higher frequencies. But that is not likely the main issue. More likely main issue is the output impedance increase near 20 kHz. When that is accompanied by drop of loudspeaker impedance, it makes causes variations to the frequency response. This is very typical for older class-D implementations. It will cause accentuated sibilance and such, in particular with LP. NAD also had a separate manual EQ compensation for this in their older amplifiers. Purifi modules shouldn't be as bad in this particular area, so worth checking out. It could be what you are looking for.

 

However, electrostatic and magnetostatic speakers are usually not the easy loads. Electrostatics also have impedance falling as function of frequency, since they are essentially just capacitors. Magnetostatics can be tricky too, so current class-D may not be best option to drive Apogee Scintilla's!

 

[davide256]

Something that also creates confusion/complexity with class D choices is the input section.  PS Audio talks about their "analog cell" input, even uses a tube buffer

in their most powerful class D amp. Apollon offers OPA 1612 as standard or offers Sparkos, Sonic Imagery OP amp choices, if you step up to their premium line

you can get Weis OP2-BP op amps.

 

Its been a while since I dabbled with op amps but my experience  then was that OP amps on a chip were just "flavors", that if you wanted more articulate sound

discrete component OP amps were what made a difference. The Hypex Nilai mentioned above says its input buffers are "fully discrete" but not sure if

that is saying the same thing.

 

Anyone able to opine on the input section and how to weigh choices in selecting a class D amp?

[Cormorant]

Yes, the Nilai input stage is fully discrete - not op amp based. 

 

The input stage is designed to work very well with the rest of the amplification circuit - it isn't a bolted on 3rd party add-on.

 

The input stage also has a shorter signal path than Purifi implementations - it's just millimeters from the rest of the circuit on the same PC board.

[barrows]

5 minutes ago, davide256 said:

Something that also creates confusion/complexity with class D choices is the input section.  PS Audio talks about their "analog cell" input, even uses a tube buffer

in their most powerful class D amp. Apollon offers OPA 1612 as standard or offers Sparkos, Sonic Imagery OP amp choices, if you step up to their premium line

you can get Weis OP2-BP op amps.

 

Its been a while since I dabbled with op amps but my experience  then was that OP amps on a chip were just "flavors", that if you wanted more articulate sound

discrete component OP amps were what made a difference. The Hypex Nilai mentioned above says its input buffers are "fully discrete" but not sure if

that is saying the same thing.

 

Anyone able to opine on the input section and how to weigh choices in selecting a class D amp?

In all of the best Class D modules (NCore, NCore, Purifi, and Nilai) the Class D stage is very close to transparent, and the first stage is what one hears.  So the choice of how the first stage is implemented makes almost all of the differences in sound (assuming the general build quality does not make obvious errors).

The NC-400 was the first Class D amp which to my ears seemed without flaw.  It used a discrete, linear input stage, powered by discrete linear regulators, all on the main module board.  Hypex's successor to the NC-400 for DIY, is Nilai, and it also uses the same approach, with a discrete linear input stage, and discrete linear regulators built into the main module board.  Now of course, just because a circuit is discrete does not mean it will sound a given way!  There are infinite circuit approaches a designer can use when going discrete, the thing is, discrete allows the designer (within the limitations of their experience and knowledge) to get the amp to perform as best as they can as they have complete flexibility in approach.

The Purifi DIY options currently available, and the commercial amps, allow for a lot of choices.  there are both IC solutions, and discrete opamp solutions available.  there are so many choices it is impossible to make generalizations about how they may, or not, sound.  One is going to have to hear them, preferably with one's own speakers.  Note that using discrete opamps, like those offered by Sparkos, Weiss, and Sonic Imagery, does not give the designer full freedom to design the circuit as they see fit, as these are integrated circuits which are what they are.  Also note that most input circuits available for the Purifi modules are simple, with one opamp acting as both the input gain source and the current driver for the Class D stage.  That topology might not be best: a better approach may be to have a first stage which handles gain, followed by a  second stage which provides the current to drive the module (this approach is used by the Neurochrome buffers designed for fronting the Purifi modules).  Also note, that a single 1612 Opamp, as used by the Purifi EVAL board is being pushed pretty hard in this application, operating near its max output capabilities for current.  By dividing current and gain requirements between two stages, a designer might be able to produce a better input stage.

Currently I am running the Purifi modules, with the EVAL 1 board.  I also have the EVAL 1 board powered by a separate linear power supply (custom) instead of the SMPS.  I also changed the onboard opamps from the 1612, to the 1656, as was suggested as an option which might be "better" by Bruno Putzeys.  I do like the sound better with the 1656: this opamp has a little more noise than the 1612, but features a higher current capability, which I think is important for driving the module.  There are lots of choices, and they all will sound a little different.

I still have not seen a commercial Purifi design which has done a fully discrete, sophisticated, input stage which addresses all the possible issues, of the IC based designs I think the Boxem looks very interesting, as they made a more complex input buffer addressing some of the issues of more simple designs.

 

 

[ericuco]

I am currently using a Rogue Audio Dragon amp which has tube (12AU7) inputs heading into Hypex Ncore modules (top right).

 

[Zauurx]

I use with pleasure a BoXem Arthur 4215 (Purifi) which is equipped with OPA1656.

https://boxem-audio.eu/en/stereo-amplifiers/10-arthur-4215e2.html

It is now connected to a Gustard A26.

No worries noted in the high register even though I always have the 20khz crossover activated in HQP for safety. This amp is perfectly built with a design beyond simple kit assembly. 👍

 

[davide256]

On 3/9/2023 at 10:25 AM, Miska said:

 

LP will spit out lot of harmonics beyond 20 kHz, because the cartridge THD near 20 kHz is pretty high. It already begins to increase from about 6 kHz. Probably not high enough to cause aliasing in the amp, but enough to cause increased power draw at higher frequencies. But that is not likely the main issue. More likely main issue is the output impedance increase near 20 kHz. When that is accompanied by drop of loudspeaker impedance, it makes causes variations to the frequency response. This is very typical for older class-D implementations. It will cause accentuated sibilance and such, in particular with LP. NAD also had a separate manual EQ compensation for this in their older amplifiers. Purifi modules shouldn't be as bad in this particular area, so worth checking out. It could be what you are looking for.

 

However, electrostatic and magnetostatic speakers are usually not the easy loads. Electrostatics also have impedance falling as function of frequency, since they are essentially just capacitors. Magnetostatics can be tricky too, so current class-D may not be best option to drive Apogee Scintilla's!

 

Magnepan included two 1 ohm 10 watt resistors with 1.7's for  tweeter attenuator jacks, using them bumps the tweeter path resistance up to 3 ohms

which has helped reduce the treble irritants from the GaN mono blocks. Its still there to some degree but better. Yea!

 

Phono cartridges don't have to have an inherent THD problem at 20khz.  I have observed that the atrophy of stylus manufacturing scale has caused

bushing mounted elliptical stylii to dominate for cartridges below USD $500 vs the sub $50 price in the 80's.  A nude mounted

hyperelliptical or other better geometry stylus that provides clean sound tracking above 10khz should take care of most of that issue,

more reason to upgrade to something like the shibata tipped Hana SL from the elliptical tipped Dynavector 10x5

[davide256]

Touched so far on 2 parts of a class D amplifier, the input circuit and the class D amplifier board but what about the 3rd, the power supply? After living with a heavy toroidal transformer + big caps in a class AB amplifier, the SMPS like bricks that came with the GaN amps were a bit surprising. Wondering at this point, how do you know if your class D amp has an appropriate power supply to get the best out of the electronics it powers?

[barrows]

A SMPS can be just as good (perhaps even better) than a linear power supply.  We are accustomed to linear power supplies, which are relatively inefficient, and require massive transformers, and huge banks of filter capacitance to produce suitable power, but SMPS have many advantages.  Many audio designers do not have the training and knowledge required to build high performance SMPS, and building high performance linear power supplies is fairly simple, and well understood by many engineers.  Building a high performance SMPS is more complicated and requires more knowledge of high speed electronics design and handling RF concerns.  This advanced knowledge is similar to what is required in order to build Class D amplifiers which actually perform well.  It is logical to conclude that if a designer is adept enough to produce a class D amplifier circuit which sounds good, that same designer can apply the same skillset and knowledge to build a high performance SMPS.

So, we as audiophiles should not dismiss SMPS, they are just another (better, more efficient) way to build a power supply.  Like anything, it is the design implementation which matters, and their are very good SMPS used in audio products right now, like EMM Labs, Chord, Mola Mola, Hypex, etc...  How do we decide if a SMPS is well designed, hmm, well does the product sound good, without obvious problems, do the measurements show clean response in the audio bandwidth?  We evaluate it just like any other product.

[Bones13]

Atma-Sphere Class D

[R1200CL]

Maybe the best amplifier ever made 😀

https://www.thetadigital.com/prometheus/

 

No problem with treble. 

https://www.stereophile.com/content/theta-digital-prometheus-monoblock-power-amplifier