Audiophiles, generally don't like class D amps!

[GXAlan]

1) One of the benefits of class D over Class AB is that there is no crossover distortion by definition, so you're not worried about any transition from Class A to B.

2) If one makes the claim that Class D switching frequency is an audible issue, then one must also make the claim that SACD high frequency noise is an audible issue. If SACD high frequency noise is NOT an audible issue, then it's hard to see how the class D switching frequency would be audible.

We don't hear the 50 kHz peak in SACD which is -80 dB on digitally analyzed test tones from SACDs.

If your speaker driver, post-crossover, is -40 dB at 500 kHz or 600 kHz then it'd meet the same -80 dB threshold in SACDs.

If you look at Sony's own Class D designs, their switching spike is only -70 dB at 375 kHz or so. Those Class D designs have no negative feedback so they have more distortion, but the same argument applies: The DSD high frequency noise is likely to be a bigger issue than the Class D switching noise, and since DSD HF noise isn't a real issue, neither is the Class D switching noise.

L7 Audio Lab on the Walkman WM1Z:

Here are measurements of the TA-ZH1ES for DSD64 vs DSD256 up to 192 kHz.

 

[dlaloum]

1) One of the benefits of class D over Class AB is that there is no crossover distortion by definition, so you're not worried about any transition from Class A to B.

2) If one makes the claim that Class D switching frequency is an audible issue, then one must also make the claim that SACD high frequency noise is an audible issue. If SACD high frequency noise is NOT an audible issue, then it's hard to see how the class D switching frequency would be audible.

We don't hear the 50 kHz peak in SACD which is -80 dB on digitally analyzed test tones from SACDs.

If your speaker driver, post-crossover, is -40 dB at 500 kHz or 600 kHz then it'd meet the same -80 dB threshold in SACDs.

If you look at Sony's own Class D designs, their switching spike is only -70 dB at 375 kHz or so. Those Class D designs have no negative feedback so they have more distortion, but the same argument applies: The DSD high frequency noise is likely to be a bigger issue than the Class D switching noise, and since DSD HF noise isn't a real issue, neither is the Class D switching noise.

L7 Audio Lab on the Walkman WM1Z:

Here are measurements of the TA-ZH1ES for DSD64 vs DSD256 up to 192 kHz.

The logic is pure.... but the purists hate it!

 

[Bjorn]

In that case, it's a matter of philosophy when it comes to recordings. Do you just shrug at the fact that the mix isn't very good? Or do you take matters into your own hands and tweak the mix to your liking?

My system is balanced enough that it can tolerate all music. But many systems and living rooms aren't there, and I understand well why some will use electronics with audible distortion to make it more pleasant and enjoyable.

 

[nowonas]

Interesting discussion and great thread!

I see that some people explain why people prefer A/B and class A amplifiers with the argument that they like the sound of distortion. I find that argument very strange, as the best measuring amplifier on this site (at least when the measurement was performed), is the Topping LA90 class A/B amplifier. Low SINAD numbers is, the way I have understood it, more coupled with the amount of feedback in amplifiers and not with amplifier topology. Class D amplifiers usually have a large amount of feedback in order to be stable and hence impressive SINAD.

I also see that some arguments that class D designs are inferior in the middle/high range, I have to admit that is also my impression. This is also inline with the findings and implementation of the new KEF designs, which have chosen A/B for the HF bandwith. From the KEF LS60 white paper (https://assets.kef.com/product-support/ls60-wireless/LS60W_Whitepaper.pdf) :

"Class AB has the added advantage of wide HF bandwidth and of not requiring an output filter"

Can this be one explanation of why some do not like class D in the HF bandwith? That the required low pass filter with limited bandwidth cause audible ringing and phase errors in the audible range?

I know that this is not the case with modern class D amplifiers, like Hypex, which I agree sound fantastic.

 

[fpitas]

Class A does not have to be a distortion box. My J2 has 0.008% distortion where I use it.

 

[SSS]

Was reading many of the inputs to this post. Well, the sound may be OK. But I don't want to have an AM transmitter in my room. Further it is well known that after some years at switching supplies the electrolytic caps deteriorate or sometimes explode. My analog power amps are up to 40 years old without any problems and still sound good. This is the reason staying with analog and not going to class-D power amps.

 

[fpitas]

But I don't want to have an AM transmitter in my room.

Still, you may not need speakers then. You can listen through your teeth!

 

[ahofer]

his is also inline with the findings and implementation of the new KEF designs, which have chosen A/B for the HF bandwith.

I think cost and additional design/space tradeoffs are what dominated here. They needed significantly less power for the high end.

Class AB can sound indistinguishably perfect relative to class D and has its own design tradeoffs. That doesn't make it inevitably superior (or worse).

 

[ahofer]

I don't want to have an AM transmitter in my room

How do you feel about wi-fi access points?

I have to say, when people say "it is well-known" in passing, my reaction tends to be "better look that up". Pretty sure that's a design/parts quality issue.

 

[fpitas]

I think cost and additional design/space tradeoffs are what dominated here. They needed significantly less power for the high end.

Class AB can sound indistinguishably perfect relative to class D and has its own design tradeoffs. That doesn't make it inevitably superior (or worse).

Yes. I would take what they say in that paper with several grains of salt. They have some fine engineering, but some intern probably put the paper together.

 

[fpitas]

How do you feel about wi-fi access points?

I have to say, when people say "it is well-known" in passing, my reaction tends to be "better look that up". Pretty sure that's a design/parts quality issue.

Wi-fi puts out a lot more radiated power, because it is allowed to by law.

 

[nowonas]

I think cost and additional design/space tradeoffs are what dominated here. They needed significantly less power for the high end.

Class AB can sound indistinguishably perfect relative to class D and has its own design tradeoffs. That doesn't make it inevitably superior (or worse).

I am not so sure that a class A/B amplifier is cheaper to produce than class D. However, if your design goal (as KEF states) is an amplifier with high bandwith, I would guess that it is cheaper to use A/B, as it is only the premium (like Hypex) class D manufacturers that produce class D amplifiers with very high bandwidth.

 

[DanielT]

1) One of the benefits of class D over Class AB is that there is no crossover distortion by definition, so you're not worried about any transition from Class A to B.

But surely that problem, or challenge, was solved 40 or 50 years ago?

I see that some people explain why people prefer A/B and class A amplifiers with the argument that they like the sound of distortion. I find that argument very strange, as the best measuring amplifier on this site (at least when the measurement was performed), is the Topping LA90 class A/B amplifier. Low SINAD numbers is, the way I have understood it, more coupled with the amount of feedback in amplifiers and not with amplifier topology. Class D amplifiers usually have a large amount of feedback in order to be stable and hence impressive SINAD.

I also see that some arguments that class D designs are inferior in the middle/high range, I have to admit that is also my impression. This is also inline with the findings and implementation of the new KEF designs, which have chosen A/B for the HF bandwith. From the KEF LS60 white paper (https://assets.kef.com/product-support/ls60-wireless/LS60W_Whitepaper.pdf) :

"Class AB has the added advantage of wide HF bandwidth and of not requiring an output filter"

Can this be one explanation of why some do not like class D in the HF bandwith? That the required low pass filter with limited bandwidth cause audible ringing and phase errors in the audible range?

I know that this is not the case with modern class D amplifiers, like Hypex, which I agree sound fantastic.

In the end, it is the entire chain that must be investigated. Distortion and noise are not conjured away but added step by step, for each gadget you plug in:

How do you sum up distortion in the sound chain?

That's the question. Anyone have the answer? It's not as simple as just a plus, is it? For example 0,03% + 0,3% = 0,33%? Edit: After a bit of googling that it seems to be about some kind of Root Mean Square calculation ..Or...

audiosciencereview.com

Speaking of SINAD, an example. This active speaker Neumann KH420. Very good criticism, good speakers according to Amir. Neumann KH420 are probably really good (speakers I could absolutely imagine having ) but check its amplifier's distortion level, 0.1%

Neumann KH420 Review (Studio Monitor)

This is a review and detailed measurements of the Neumann 3-way, active speaker (monitor). It is on loan from the company and costs US $5,195. Speaker was too heavy to lug to the loft where my main system is so you have to put up with the cell phone picture above and listening tests there. Back...

audiosciencereview.com

____

The problem, as I understand it, with class D is intermodulation distortion, IM distortion. Distortion in the higher inaudible registers that "spills over" into the audible registers. If that happens depends on the amp design, what type of post amp filter (the last thing on the amplifier filter) there is.
Note I could be wrong! More knowledgeable EE can explain more about that. I too am curious to know more. I also wonder why Amir only used 45 kHz bandwidth in his class D amp measurements?

Edit:
Regarding Neumann KH420.Good with power though, which I think is the most important thing when it comes to amps (unless they are terribly poorly designed). Amplifiers, with little power, that are driven into clipping sound AWFUL. Think crappy old car stereo, clock radio. The sharp kind of unpleasant sound when pushing them is amp clipping (most often, distorting speakers sound "softer" in their discomfort), brrrr.
In addition to the fact that low powered amps can destroy your interest in listening to music, they can also damage, supply tweeters.

 

[SSS]

How do you feel about wi-fi access points?

I have to say, when people say "it is well-known" in passing, my reaction tends to be "better look that up". Pretty sure that's a design/parts quality issue.

Wi-fi is a totally different frequency range. And yes, a cellphone near audio gear can induce noise to it. But 500kHz and some higher can also influence the audio circuits.
Further, well-known regarding electrolytic caps does not mean that all switching supplies have this problem. But I had several of bad ones which I needed to repair. And this problem is stated in several electronic forums. Maybe today best eletrolytics with very low ESR will stay longer in switcher applications.

 

[nowonas]

The problem, as I understand it, with class D is intermodulation distortion, IM distortion. Distortion in the higher inaudible registers that "spills over" into the audible registers. If that happens depends on the amp design, what type of post amp filter (the last thing on the amplifier filter) there is.
Note I could be wrong! More knowledgeable EE can explain more about that. I too am curious to know more. I also wonder why Amir only used 45 kHz bandwidth in his class D amp measurements.

Interesting point about intermodulation distortion, as intermodulation distortion is a much more audible concern than many other types of distortion ( ref klipch distortion test earlier this thread). I would also guess that intermodulation distortion is very difficult to measure properly using only static measurements (as the industry does) as intermodulation will be more prominent in highly dynamic settings and difficult to simulate using static measurements.

 

[SoNic]

Interesting point about intermodulation distortion, as intermodulation distortion is a much more audible concern than many other types of distortion ( ref klipch distortion test earlier this thread). I would also guess that intermodulation distortion is very difficult to measure properly using only static measurements (as the industry does) as intermodulation will be more prominent in highly dynamic settings and difficult to simulate using static measurements.

That is the problem. Static measurements are only the beginning of the audio quality evaluation, not the end.
However a lot of people have developed some sorts of religious fervor in believing that static measurements are "ne plus ultra".

 

[SSS]

That is the problem. Static measurements are only the beginning of the audio quality evaluation, not the end.
However a lot of people have developed some sorts of religious fervor in believing that static measurements are "ne plus ultra".

It is right that listening is the final judgement for specific music and its listener. Here told static measurement is not static. The amplifer output swings with superseded frequencies permanently from + to - voltage values and thus currents. Specially the multi-tone IM is revealing. Also the TIM measurement exists which is as far as I found not much done anymore. By the way, intermodulation is much higher in almost all speaker chassis than in amplifiers when driven to some relevant SPL.

 

[Kijanki]

I'm not sure why class D has to have, as often stated, "large amount of feedback" to be stable. If anything, amplifier can be unstable because of the feedback. Output stage is "unstable" to start with, since output Mosfets switch from open to close. It cannot be more "unstable" than that. In addition it has low output impedance without feedback, since one of the Mosfets is always shorted to VCC or GND - no need for a lot of feedback. As for limited bandwidth - yes, as it is now it is limited to about 50kHz. One can argue, that it causes about 20deg phase delay at audible 20kHz hence improper harmonics summing, but speakers at this frequency are usually inductive, accelerating phase. If frequency limit to 50kHz is audible then why in Stereophile review of Lamm ML3, that has the same 50kHz@-3dB limitation, they stated "Most startling, though, was the ML3s' effortlessly clean and airy high-frequency extension" (AFAIK there is no way to limit bandwidth without phase delay). I had impression that treble do sound different with my previous class D amp - cymbals sounded slightly more natural - brassy and not splashy. It was early Icepower and now my Benchmark's AHB2 sounds better, but not by much. Class D is even better these days.

High IMD? Not according to measurements. It doesn't surprise me since quantity of interest "duty cycle" is more linear that output voltage of class AB amp and non-linearity is the cause of IMD. It is also inconsistent with "large amount of feedback" since negative feedback improves linearity.

I've also heard that about 1% of the carrier residue can result in modulation on the non-linear tweeter. True, but in order for this to happen tweeter membrane has to move and it won't happen at 500kHz for sure.

I remember well known reviewer making a lot of negative comments about class D long time ago. At one point I realized he thinks it is a "sampled system" (limited resolution). People not opened to something new will always find something wrong and will even hear it (negative placebo effect). That's life.

 

[Ricardus]

I've been on ASR for less than 2 years, but the point has been made almost endlessly by everyone from Amir down (or out, to be fair to our best minds) that not all playback attributes are measurable, and that controlled listening tests are the final tie-breaker.

It has? I almost never see people on here touting listening tests.

 

[NTK]

That is the problem. Static measurements are only the beginning of the audio quality evaluation, not the end.
However a lot of people have developed some sorts of religious fervor in believing that static measurements are "ne plus ultra".

Quoting Doug Self:

"Sinewaves are steady-state signals that represent too easy a test for amplifiers, compared with the complexities of music."​

This is presumably meant to imply that sinewaves are in some way particularly easy for an amplifier to deal with, the implication being that anyone using a THD analyzer must be hopelessly naïve. Since sines and cosines have an unending series of non-zero differentials, "steady" hardly comes into it. I know of no evidence that sinewaves of randomly varying amplitude (for example) would provide a more searching test of amplifier competence.​

I believe this outlook is the result of anthropomorphic thinking about amplifiers; treating them as though they think about what they amplify. Twenty sinewaves of different frequencies may be conceptually complex to us, and the output of a symphony orchestra much more so, but to an amplifier both composite signals resolve to a single instantaneous voltage that must be increased in amplitude and presented at low impedance. The rate of change of this voltage has a maximum set by the frequency response and amplitude capability of the channel and is not generally greater for more complex signals; you do not get higher slew rate with bigger orchestras. You must remember that an amplifier has no perspective on the signal arriving at its input, but literally takes it as it comes.​