Class D: Turns Out it Does Suck!

[GUTB]

Taken second-handed from someone who posted the following from an Absolute Sound Roundtable of master amp designers on class D:

 

Bob Carver
"I built many of them right here in my own laboratory with the thought they could and would fulfill that final promise.... I was never able to build a Class D amplifier that sounded as good as a linear one."

 

John Curl (Parasound, CTC, Vendetta Research, Constellation)
"Some version of hybrid Class A/D looks like the future in optimum audio design."

 

Cyrill Hammer (Soulution Audio)
"If you want to have your product performing at the cutting edge it is not possible with today's known switching technologies. In order to come close to the performance of the best linear design we would need high-current semiconductors that provide switching frequencies of several MHz or even GHz."

 

Lew Johnson (Conrad Johnson)
"I tend to think that Class D circuit design is an approach best relegated to producing low-cost, physically manageable multichannel amplifiers where one might accept some compromise in sound quality for the sake of squeezing five, six, or seven 100 watt channels into one moderate-sized package for a budget home-theater installation."

 

Vladimir Shushurin (Lamm Industries)
"No, it is not. And I would like to respond to the second part of this question with an allegory. Any field of human activity defines a number of requirements which, when properly implemented, guarantee a positive outcome. For example, the basic requirement in the army and sports is an able-bodied individual. So, it would be quite natural to concentrate on searching for such an individual (especially as we know where to find him). However, out of the blue we decide to choose a feeble-bodied person who, on top of that, is encumbered by various diseases. Having made this decision (which is a priori improper) we start justifying it to ourselves and others by citing the great state of our medicine, which is capable of curing many ailments."

 

(Savage! -- GUTB)

 

Fumio Ohashi (BAlabo)
"No. Class D can't really be considered for super-high-end performance in its present stage of development, although it can be fine for mid-market products."

 

Nelson Pass (Pass Labs)
"Does a $10 bottle of wine compete with a $100 bottle? Of course it does, and it often wins based on price. Right at the moment Class D designers seem to be still focusing on the objectively measured performance of their amplifiers. I expect that at some point the economics of the marketplace will encourage them to pay more attention to the subjective qualities, and then they will probably play a greater role in the high end."

 

(Most diplomatic -- GUTB)

 

Jürgen Reis (MBL)
"I have worked a lot lately with Class D. Ninety-nine percent of Class D circuits are not competitive with linear circuits. Most Class D sounds sterile. It's tricky to figure out what to do to compensate for that."

 

Thorsten Loesch (iFi - AMR)

"I have yet to hear a pure class D Amp I’d rate above "below average for solid state" (which is not very high performance). In a little update of my classic 'Valve Analogue Stages for DACs' I wrote: "Perhaps more crucially, so called Class D Amplifiers, which have in recent times sprouted up like mushrooms after a warm rain, continue to use the straight two or three level modulation scheme described above. And thus they still require the use of heavy handed noise shaping to attain anything like acceptable 16 Bit Audio performance.The clock frequencies for these amplifiers are usually at 300 KHz to 1MHz in the best cases. That is 3,000 to 10,000 times lower than what is required to attain 16 Bit / 44.1 KHz performance without noise shaping and other forms of signal manipulation! And again, one is baffled and perplexed by the rave reviews many Class D amplifiers receive, as baffled as one was about the late 90s reviews of timeslicing DACs. The best of breed I have auditioned were certainly not bad; however in direct comparison to the best available valve and solid state amplifiers they do not produce a very good sound. Well, at least they offer novelty and the reviewers something to write about other than another (however good sounding) 8 Watt valve amp. Incidentally, the best sounding Class D amps tend to be really low power single chip devices (putting out little more than the 8 watt valve amps), presumably because they are faster AND because they always work near what one might call a full scale, if they would be DACs. On second thought, they of COURSE are DA Converters and where a Class D amplifier accepts analogue input directly it is an A2D converter followed by a power D2A converter! What an insight!?"

 

Mark Levinson

"Interleaving of multiple Class D Amplifiers is potentially a step in the right direction, but does not go far enough.Personally I think that the best option would be something that combines a Class D Amplifier for the heavy lifting with something Class A for fine detail. Probably implemented in the style I did for AMR’s AM-77 "Jikoda" Style. In this case both of the circuits involved can operate fully open loop. In many ways the problems in Class D Amplifiers are analogous (but not identical to) those in Class B Amplifiers (but without an option to implement Class AB or Class A) so similar solutions apply. All Class D amplifiers are essentially delta-sigma DAC’s. If the input is not digital PWM signals (aka "DSD") but analogue audio then it is also a Delta Sigma Analogue to digital converter...Now DSD (aka SACD) which to my ears fails to come close, never mind equal true PCM CD Replay in most aspects of sound quality, operates at 2.8MHz switching, or around 10 times as fast as common Class D Amplifiers...Why anyone would want to listen through an A2D followed by an D2A Converter that are around 10 times worse than single speed DSD is beyond me. But with enough hype and snazzy naming it cannot help but sell high and wide." 

 

And, don't worry, we haven't forgotten:

 

Jeff Rowland
"I consider Class D to be highly competitive in the present, and to offer an evolutionary pathway of audio design that may produce even more astonishing results in the future."

[GUTB]

4 hours ago, wgscott said:

 

Devastated!!

[GUTB]

6 hours ago, firedog said:

First, another silly attempt of yours to troll, isn't it?

I will partially take the bait:

 

1. This is mostly comments of guys who don't do Class D. Not exactly an objective group. In addition, designing Class D properly takes specialized knowledge and experience if you want good results. Probably lots of these guys, even thought they may be great designers, haven't given the time or thought to Class D to make it work, and don't truly understand it on the most sophisticated level.

 

2.  This is only a few years old, but already dated. Several newer generations of various Class D technology have come out in the past few years. General consensus is that they ARE competitive with non-Class D solutions, and often thought to outperform them, dollar for dollar, except at the most expensive, over $10k levels. 

 

I'm sorry if quoting legends in amp design is "trolling", but I don't believe there have been any fundamental improvements in class D since then....has there? Which specific modules are you talking about?

[GUTB]

1 hour ago, blownsi said:

While I realize that all of those guys have far more experience and expertise than I will ever have, none of them have my ears.  To my ears Class AB "sucks" and I have no idea why they bother with it at all.  Build a proper Class A like Mr Pass or imitate one with a quality Class D design is what sounds the best to me.  

 

I've heard a couple of class A and A/B amps which suck. But the problem with class D is that it sucks by nature. That's not a troll, but a hard fact. Vladimir Shushurin from Lamm was the most brutal about the issue, calling class D feeble with a bunch of diseases on top of it, while we deal with all of its shortcomings with modern medicine.

 

Guess which one of these 10 KHz square waves is from a class D and and which one is from a class A amp:

 

 

 

[GUTB]

21 minutes ago, firedog said:

And it proves absolutely nothing about Class D amps as a class. Your so called "fact" is an opinion, not a fact. 

 

We get it, you don't like ClassD. 

 

What you don't get is that nothing you think is provable. Every argument you are going to make can be countered - believe me, we've already heard them all. The arguments aren't new. 

Like the rest of us, you are free to think what you want and come to whatever conclusions you want. 

What I don't get is your compulsion to try and prove it and argue about it.

I think that defines "troll"....

I'm out of this one, I can tell it will be a useless so-called "discussion".

 

Would you care to explain why these legends of amp design are wrong?

[GUTB]

2 minutes ago, kilroy said:

Ignore the troll.

 

Hey, here is what several top amp designers think.

Check out these screwed up measurements.

 

"Ignore the troll". Lol.

 

Look. I've listened to various class D products including the latest Pascal module. It's true modern class D has come a long in my experience from "unlistenable" to "excellent with a few deadly flaws". But, in the end of the day, it doesn't matter how resolving and controlled it is if its filters destroys soundstage depth and it struggles with musicality. I wish I could just buy a sub-$2000 amp that I can pick up with one hand and be happy with it. I'm sorry I have certain expectations from speaker listening.

[GUTB]

Well, as a matter of fact, I'm still resisting upgrading to an end-game SET. I have a few options on the table, the one I'm leaning towards now is an enormous pair of Line Magnetic 845 monoblocks that weigh in at 140 lbs and cost $6,000 from China. I'm also going to the Dallas audio fest this weekend where I will be checking out KR Audio among others. But I still don't WANT to have big iron in my listening room.

 

I heard the Linnenberg monos at AXPONA and was quite impressed with the sound. I'm arranging an in-home demo (ie, purchase with return option lol) with their US dealer as we speak. These are super high-bandwidth (350 KHz after low pass filter), very high-speed, low-noise amplifiers at "only" $5,500. I'll try these out.

 

Even though the weight of evidence does seem to suggest class D sucks, I'm still listening to counter opinions.

[GUTB]

53 minutes ago, ShawnC said:

A local dealer has Bel Canto's Black series The system (class D mono amps) (expensive) hooked up to Wilson's Alexia's speakers.  It's the best sounding system I've heard them assemble.  A while back they had the now discontinued Audio research DS450 (class D amp) and Magnepan 3.7 together. That also sound amazing.  The dealer said that the DS450 was there best selling amp.

 

 

 

I listened to a Vandersteen Channel D combo, it was basically unlistenable but I don't necessarily blame that one the amp.

[GUTB]

Genelecs use class A/B

 

[GUTB]

6 minutes ago, Fitzcaraldo215 said:

Another meaningless anecdote.  I had been using mid-90's vintage Krell KAS-2 monoblocks in my system, exceeded only by the KAS-1 at the very top of their lineup at that time.  Class A, of course.  KAS-2 MSRP then - $22,000 the pair.  

 

About 7-8 years ago, I auditioned against the Krells in my own home a bunch of different hi quality amps  - classes A, A/B and D.  I did not like the Bel Canto or other B&O ICE-based Class D's back then, for example.  But, my favorite amp by a fair bit was a little-known, direct from the mfr. Spectron Musician III Mk. 2 Class D, which I thought sounded better than all of them, including my Krells.  

 

I sold the Krells, and I am still the very happy owner of the Spectron. $4k MSRP, by the way.  Spectron's founder, John Ulrick, died about a year or so ago, though the company goes on. He developed the first switching amp in  audio history when he worked for Arnie Nudell at the old Infinity, when it was a primo audio company.  Quite possibly, though, the Spectron is now outclassed by much newer and better sounding Class D's.

 

My take is amp class matters not at all sonically. Implementation matters much more.

 

Saying "amp class matters not at all sonically" is a position that seems really hard to justify when class D amps output waveforms like these:

 

 

Note that Stereophile stopped doing those measurements without using the Audio Precision low-pass filters to produce a waveform that looks less embarrassing. But even with the cheating, we can see class D just can't seem to handle it.

 

That's because the signal that goes INTO class D is NOT the signal which comes OUT. What comes out of class D is a modulated and filtered facsimile. That's why it has been pointed out that class D is mathematically, physically inferior to linear amplification. Of course, there are benefits to this trade-off, as always -- because class D doesn't amplify the input signal, it's also not amplifying unwanted components of the input signal. 

 

Then there's bandwidth. While the switching frequency of the NC1200 is 500 KHz. Meanwhile, Krell Evolutions have a usable bandwidth of 300 KHz...Goldmunds go up to 3 MHz (+/- 3dB) ... not even on the same planet.

[GUTB]

4 minutes ago, Fitzcaraldo215 said:

Ok, so you show a measurement of a mystery class A amp of unknown provenance and you insist they all do that, i.e., they are all the same, they are all no damn good.  They can all be judged simply,  just like people can by the color of their skin or their ethnicity. Class A = bad and you have the proof! 

 

So, you have your mind made up.  Why then are you jerking us around? Go for tubes, man.  Or, blow your money on Class A and pay the monthly electric bills.   Ain't it great that we have choices? While you are at it, you'd better start stepping back big time from all digital audio, because the output from that is a only a "modulated and filtered facsimile" of the analog input. 

 

By by the way, if you know of an amp where the input us the same as the output,  go for it! Except, no one else would call such a zero gain device an amplifier. Also, if you know of an amplifier that adds zero distortion and noise while it adds gain, go for it, man.  You have got a winner.  We mere mortals here do not know of such a device.

 

My dog and all the bats in the neighborhood also gotta have that 3MHz frequency response, which no mikes pick up, is not on any recording and which no speaker reproduces.  But, hey, if it works for you, knock yourself out.  I am quite sure you can hear the difference, unlike the rest of us.

 

 

 

As usual, no actual counter-argument, just class warfare.

 

The measurement is from the NAD M22. But please, find a single class D amp that Stereophile measures that looks much different. They used to show measurements with the low-pass filter off, which produces waveforms like this:

 

 

Anthem Statement M1.

 

Amplifier bandwidth is beneficial to the audio band.

[GUTB]

24 minutes ago, Ajax said:

From Wikipedia:

 

".... a troll is a person who sows discord on the Internet by starting arguments or upsetting people, by posting inflammatory messages in an online community with the intent of provoking readers into an emotional response,^[3] often for the troll's amusement."

 

When you reread the title of your thread "Class D; Turns Out It Does suck!" in conjunction with the comments from the amp designers, the above statement is a good fit. Inflammatory headlines is the oldest trick in the book for selling newspapers, especially when the editor is scratching for some worthwhile news or an interesting story line. It is the reason why Trump's mantra of "Fake News" has gained such traction with the public.

 

Several of the designers you quote actually state that a combination of A & D class amplification has great potential. It is still early days. Just look at the development in DAC technology over the past 25 years since the advent of digital audio. 

 

I have a Devialet 200, which parallels a low-wattage Class-A amplifier with a high-power Class-D amplifier dedicated to providing current, in a hybrid arrangement, which one reviewer stated "delivers the sweetness and harmonic complexity of solid-state Class-A circuits, with the bass power and control of good Class-D designs." To my ears, and many many others, it sounds wonderful.

 

I also have a Nord Hypex NCore NC500 amplifier driving a pair of ATC STM 19 speakers. Does it with ease.

 

FWIW, here's some advice. Making ridiculous assertions to attract attention, complete with "exclamation mark!", will loose you credibility very quickly,  There are many here at CA with greater knowledge and experience than you so if you want to gain respect on this forum I suggest you read more and write less and cut out the "look at me".

 

I suggest you take 24 hours to think about what I have written before pressing reply.

 

Thank you for the subjective review of the Devialet. Class D is definitely not in "the early" days any longer.

 

I stated that it turns out that class D sucks. To support this position I quoted some of the greatest living amp designers. It's an appeal to authority, but as I am not an authority myself I think that's okay! Calling class D "just as good" as linear circuits is actually the fake news. Alternative facts.

 

Nelson Pass diplomatically stated that a $10 bottle of wine can compare with $100 wine -- especially considering the price. It's a nice way of saying, NO, class D isn't competitive in the high end.

 

I think there is an unfortunate class warfare aspect to this topic. People with their DIY amps and sub-$2k amps want to believe that they compete with linear amps that cost many times more.

 

 

[GUTB]

10 hours ago, firedog said:

 

Hilarious that you bring up the M1 as a counter. Do you bother to read the reviews in stereophile also? Here's what the reviewer, Kal Robinson had to say in his conclusion
 

 

 

And You (of course) don't refer us to the actual review of the NAD M22 you are slagging (also by Kal Robinson), which concludes:

 

 

 Or JA's conclusion of his measurements report on the M22:

 

 

Let's see, who would I rather believe:  a troll like GUTB, or an extremely experienced reviewer like Kal Robinson, who participates here and actually tries to be helpful?  And KR,who has spent a serious amount of time reviewing the two specific amps GUTB brought to "prove" his claim that class D sucks. KR has  probably evaluated more amps of all classes than GUTB has actually heard. 

 

I posted that measurement because it's one of the only class D measurements that Stereophile put out without the low-pass filter. All class D amplifiers will have square waves that look like that.

 

In regards to the scarecrow argument about M22 subjective review impressions -- yes, I'm very impressed with the standard class D review boilerplate. "Just Music", "Super Quiet", "Resolving", etc are present. Sound stage depth and timbre rightness is, as usual, not a part of the review regimen. What professional reviewers hardly ever do is compare low-end products (lots of people might also consider the M22 to be mid-level) to high-level products. They do this with all equipment reviews, not just class D. However, sometimes you can get a comparison, like when 6moons compared a D-Sonic to Luxman monoblocks and he had to admit he preferred the Luxmans.

[GUTB]

SETs start to lose it in the upper frequencies. Pentode designs do much better. People gravitate to SETs because they have to-die-for midrange, which is the most important part of the audioband.

 

SETs, especially directed-heated "real" SETs like 845, 211, 300, etc do require proper system matching, but once dialed in they are basically unbeatable in terms of timbre rightness, harmonic richness, euphonic glow and soundstage.

[GUTB]

Since you guys are talking about these Crown amps, let's take a look inside:

 

 

Compared to an NC1200:

 

 

It's widely accepted that nCore is at the top of the class D game. If even nCore modules are unable to impress at the high end, how much can we really expect out of a $300 amp using a generic / unknown class D circuit?

[GUTB]

So, what it looks like is that class D is inferior (sound quality-wise) by nature.

 

Class A takes an analog signal and makes it larger. It does nothing more.

 

Class D takes an analog signal, converts it to a 400-500 KHz square wave, and then converts the square wave back into an analog signal while applying a filter to get rid of the high-frequency noise. What goes in, isn't what comes out.

 

The fundamental problems of class A are: efficiency and noise.

 

The fundamental problem of class D is: low sound quality.

 

Class A will probably never be able to solve its fundamental problems. It has to run at at full power all the time, pouring energy into un-used cycles constantly. Anything less, and it's no longer class A.

 

But can class D solve its fundamental problem? It seems likely! More importantly, has class D solved it's problem already? You guys with class D, don't be shy. Talk about your class D amps, your setups, the sound you get from them and how they compared with other amps.

 

[GUTB]

19 minutes ago, Jud said:

 

Since you've successfully fended off all attempts at educating yourself so far (i.e., what you've written is exactly what you were saying to start the thread), why should anyone bother to make the effort?

 

I'm sorry, but it seemed to me I was the one doing the educating. In return, I was "educated" with: "Troll", and "You're wrong", and "class D is great".

[GUTB]

49 minutes ago, Ralf11 said:

you're not backing up your conclusions about Class D

 

Let's forget about mechanism, and just explain which Class D amps you've heard, the cost, and how old the design is with some specifics on the particular SQ flaws each time

 

Okay. Amps I've owned:

 

Several home theatre receivers....won't pick on them too much no one should expect much from them regardless of class.

 

Teac AI-301DA. Uses an ICE 50ASX2-SE. Basically unlistenable. $350.

 

D-Sonic M3-800S. Uses a Pascal module (latest series not sure which) and a custom input stage. Resolving, controlled, fast, dynamic. Lean on bass. Soundstage depth crushed. $1,400.

 

I've listened to several others at AXPONA, but I can't in fairness judge them based on that experience due to the poor mess of acoustics, bad system tuning, noise and bustle of crowds.

[GUTB]

2 hours ago, Jud said:

 

And the DAC conversion.

 

Our thread host has managed to prove he never heard decent audio except from a turntable, tape deck, or live performance, because all digital audio works in exactly the way he says is irreparably bad.

 

Class D modulation is similiar to DSD -- 1-bit square wave. Except that single-rate DSD suffers from steep ultrasonic noise escalating from 25 KHz, and operates at 2.8 MHz. Class D switches between 400 and 500 KHz, and suffers from massive ultrasonic noise that wreaks havoc on speakers unless filtered. PCM operates at much lower frequencies except it's also at least 16 bits wide. Notice that that the slowest clock you will find in any DAC is 2.8 MHz -- do you guys know why that it?

 

Basically, class D modulation is a super low-resolution ADC.

 

 

[GUTB]

3 minutes ago, Fitzcaraldo215 said:

OK, so he listens to exactly two cheapo examples and he concludes that therefore ALL class D is junk. Does anyone else see a flaw in this logic?  

 

There are morons, but then there are total morons.  Unfortunately, some of the total morons wish to make a flagrant, public spectacle of themselves for all to see.  Mind-boggling, and, unfortunately, of totally no use on the question of the audible advantages/disadvantages.

 

The D-Sonic uses one the latest class D modules and is unverisally well-reviewed. $1,400 is in the $1,000 - $2,000 price range for modern class D implementations using NC500 and similiar technology. It cannot be considered an el-cheapo model. Only NC1200 products are greatly more expensive. Jeff Rowland has a $60k product. There's a $15k Technics using the new GaN FETs which would be interesting to listen to.

[GUTB]

Just now, Ralf11 said:

While I am sorry that your soundstage was crushed and your money wasted, it does seem that this event has overactivated your limbic system.

 

I returned the product after 30 days, so I only wasted a few bucks. It was a great learning experience too so I can't call it a waste of time either.

[GUTB]

5 minutes ago, Jud said:

 

Since you've said you're the one doing the educating in this thread, I'm sure you know far more about this than Bruno Putzeys, designer of one of the world's best ADCs used by audiophile labels.  So please educate us and Bruno about how to make a good ADC.

 

I don't know about his ADCs.

 

Can someone find the the quote where Bruno Putzeyes says that class d sounds good in SPITE of being class d?

[GUTB]

2 minutes ago, Fitzcaraldo215 said:

D-Sonic "universally"  well-reviewed? What reviews would that include, exactly?  

 

My friend, you are, among other obvious faults to your thesis, just not at all well educated on the available products, duh!  But, little  things like that should not limit you from proclaiming from the mountaintops that you have discovered the ugly and unfortunate truth about Class A, that you know all about this technology and that you can objectively lead us all to reality.  

 

Dream on.

 

Which class D amps impressed you?

 

Do current class D modules switch at 400-500KHz? Does it not matter?

[GUTB]

I listened to to a Linnenberg Allegro monoblock setup at AXPONA that impressed me. I walked over to the price sheet expecting to see $10-15k, it was "only" $5,400. I arranged for an in-home demo which I expect to receive today. The amp is class A/B, features a very high-speed, high-bandwidth design (350 KHz after filter). Very high peak current draw for the size/price. Very low noise for class A/B (-116 dBA, non-weighted). It'll be interesting to compare the sound through my speakers compared to the class D products I've owned.