Class A vs AB -- Do They Really Sound Different?

[Doodski]

There's something great about having an amp that is full out at idle waiting for the slightest signal to output. Nice warmth in the room and if linear then that is just icing on the cake. Class A @ 75R8 150R4 300R2 full out 1Khz p-p driven into loads was the stock spec. I never noticed anything out of spec whenever I took 'em out for a good test. I don't think the class A was the only reason the amp sounded so good. It's linearity was.

 

[avanti1960]

I don't trust my ears.

sorry to hear that
when i hear class a amplifiers or class ab amplifiers that remain class a at higher power than typical i hear a smoothness to the sound that is a consistent, tell tale characteristic. subtly but noticeably different than most class ab amps.
however, whether it sounds better is debatable. some class a amps, even some very expensive, sound too smooth for my preferences and seem to diffuse or veil musical details. best partenered with revealing speakers imho.
my current amplifier stays class a up until 12 watts and it is definitely smoother sounding than any other amplifier i have owned.
resolution is slightly diffused but my speakers are exceedingly clear sounding and the overall sound is very nice. fyi i listen only to music, no HT.

 

[watchnerd]

sorry to hear that

What's to be sorry about?

After having done a lot of level-matched ABX testing, I'm just acknowledging my limitations and propensity for sighted bias.

 

[MakeMineVinyl]

I'd bet pentode/beam tetrode based tube amps have a whole bunch more global NFB, just by nature of both types having dramatically more available gain - I've not checked the loop gain of any myself, but it's a hunch.

You're undoubtedly right but single ended triode amps don't use tetrodes or pentrodes. That's the point of their existence.

 

[SIY]

I'd bet pentode/beam tetrode based tube amps have a whole bunch more global NFB, just by nature of both types having dramatically more available gain - I've not checked the loop gain of any myself, but it's a hunch.

Depends on the topology. For example, Circlotrons have pretty low "global" feedback. (scare quotes intentional) Ditto Unity Coupled.

 

[dualazmak]

Uh huh.

Hello again SIY,

How do you think about SONY's descriptions on TA-A1ES, especially;
"The TA-A1ES uses a current feedback type power amplifier to achieve a sound with excellent focus of the sound image without phase rotation up to the ultra-high range. "
(This looks like similar to Accuphase's description for A-36 as shared here.)
and
"In addition, by applying sufficient bias, the heat generation of the transistor becomes almost constant, there is no sudden temperature change at the semiconductor chip level, phase rotation in the ultra-high range is suppressed, you can enjoy a sound with good focus, and the emitter resistance is deleted. "

In any way, I am planning to test and evaluate TA-A1ES for Beryllium Tweeter and Horn Super Tweeter in my multichannel multi-amplifier project.

 

[SIY]

Hello again SIY,

How do you think about SONY's descriptions on TA-A1ES, especially;
"The TA-A1ES uses a current feedback type power amplifier to achieve a sound with excellent focus of the sound image without phase rotation up to the ultra-high range. "
(This looks like similar to Accuphase's description for A-36 as shared here.)
and
"In addition, by applying sufficient bias, the heat generation of the transistor becomes almost constant, there is no sudden temperature change at the semiconductor chip level, phase rotation in the ultra-high range is suppressed, you can enjoy a sound with good focus, and the emitter resistance is deleted. "

In any way, I am planning to test and evaluate TA-A1ES for Beryllium Tweeter and Horn Super Tweeter in my multichannel multi-amplifier project.

I don't pay much attention to ad copy. Especially when translated from Japanese marketing to English. The amplifier either makes a small signal bigger with little alteration (the vast majority of engineered products) or it doesn't. No magic there.

 

[dualazmak]

I don't pay much attention to ad copy. Especially when translated from Japanese marketing to English. The amplifier either makes a small signal bigger with little alteration (the vast majority of engineered products) or it doesn't. No magic there.

Maybe your policy is somewhat same as mine; "Trust our ears and brain" in actual settings, right?

 

[SIY]

Maybe your policy is somewhat same as mine; "Trust our ears and brain" in actual settings, right?

I trust my ears. Being a human, I realize my brain lies constantly so my perceptions need to be checked.

 

[MrPeabody]

I specifically use class A amplification on my high frequency horns because of crossover distortion in class A/B. Don't ask me to prove it with graphs and charts because it's none of your business.

Strange. If you didn't want anyone to ask you to prove the claim, why did you come here and wave it in everyone's face?

 

[MrPeabody]

The only question here that is open to debate is whether the crossover distortion of a well-designed Class AB is audible. If the answer is decidedly "no", then this immediately becomes a silly debate. Anyone who wants to claim the superiority of Class A needs to offer evidence of the audibility of distortion in Class AB amps, not cherry-picking some poor example, but ideally showing that even in the Class AB amps that measure among the best of the breed, crossover distortion remains audible. And we should all know by now that to do that, it is necessary to prove it using properly controlled blind testing protocols. Of course there will always be a few people who will immediately perk up and offer the exceedingly tiresome argument that blind testing is worthless if doesn't agree with what they "know" that they hear. When considering the question of audibility of distortion of Class AB, I would maybe pay some extra attention to distortion at low volume. I don't know ... it's the only thing I can think of where there might be a possibility that Class A has an advantage. I'm sure that there are people here who have a very good understanding of this question and should be able to shed some light on how THD + S/N breaks down into THD and S/N at low playback level, and whether there is an inherent difference between Class A and Class AB with respect to accuracy at low playback level.


It also occurs to me that there are other similar questions where an open discussion might be more informative. One example I can think of is the question of whether "soft clipping" is commonly implemented in amplifiers, and if so, whether it is particularly likely that actual distortion levels may be appreciably higher than people generally take for granted, owing to either (a) masking of distortion by the soft clipping, or (b) the soft clipping itself.

 

[watchnerd]

The only question here that is open to debate is whether the crossover distortion of a well-designed Class AB is audible. If the answer is decidedly "no", then this immediately becomes a silly debate. Anyone who wants to claim the superiority of Class A needs to offer evidence of the audibility of distortion in Class AB amps, not cherry-picking some poor example, but ideally showing that even in the Class AB amps that measure among the best of the breed, crossover distortion remains audible. And we should all know by now that to do that, it is necessary to prove it using properly controlled blind testing protocols. Of course there will always be a few people who will immediately perk up and offer the exceedingly tiresome argument that blind testing is worthless if doesn't agree with what they "know" that they hear. When considering the question of audibility of distortion of Class AB, I would maybe pay some extra attention to distortion at low volume. I don't know ... it's the only thing I can think of where there might be a possibility that Class A has an advantage. I'm sure that there are people here who have a very good understanding of this question and should be able to shed some light on how THD + S/N breaks down into THD and S/N at low playback level, and whether there is an inherent difference between Class A and Class AB with respect to accuracy at low playback level.


It also occurs to me that there are other similar questions where an open discussion might be more informative. One example I can think of is the question of whether "soft clipping" is commonly implemented in amplifiers, and if so, whether it is particularly likely that actual distortion levels may be appreciably higher than people generally take for granted, owing to either (a) masking of distortion by the soft clipping, or (b) the soft cliHI pping itself.

I can't figure out a way to do an ABX listening test for this using ordinary consumer electronic equipment that wouldn't also introduce a crap ton of other variables that may affect the outcome.

Wouldn't you probably need a specialized testing rig?

 

[levimax]

A typical tube SET class A amplifier doesn't have a whole lot of loop gain to begin with. Mine have about 15dB of overall gain, of which I'm using about 13dB for global negative feedback - that leaves the amp with only 2dB of gain which is perfectly fine with my HF horns which have something like 107dB for 1 watt efficiency. The thing is, components have to be selected to fit into an overall system. A single ended class A amp would be stupid to the extreme to use as the main amplification for powering a moderate to low efficiency speaker with traditional passive crossovers. The vast majority of people have conventional speakers like this, so in a way the premise of this thread is kind of a moot point. People who need low power class A amplifiers know who they are, and they will seek them out. The rest of users are best served by conventional higher powered class A/B or D amplifiers.

In other words, move on, there's really nothing to see here.

What do you think you gain with a Class A Amp? I used one on the tweaters of a tri amp system and then one summer tried a Neurochtome amp to save on AC and besides being cooler , smaller, and quieter I could not tell any difference so I stuck with the AB amp.

 

[SIY]

The only question here that is open to debate is whether the crossover distortion of a well-designed Class AB is audible.

Or if it even exists in detectable quantities in engineered products. I don't see it in any of the low power or high power traces from any of the AB amps that have cycled through here in the past few years.

Self has some good analyses and data on this.

 

[dfuller]

The only time I've ever seen crossover distortion is when a guitar amp (which is decidedly not a high fidelity amplifier) is significantly overbiased. I've never once seen it in a high fidelity amp assuming proper bias.

 

[MrPeabody]

I can't figure out a way to do an ABX listening test for this using ordinary consumer electronic equipment that wouldn't also introduce a crap ton of other variables that may affect the outcome.

Wouldn't you probably need a specialized testing rig?

I don't know. I would hope that the switching equipment needed to do proper ABX testing would be simple enough that this wouldn't be a concern. But I didn't mean to suggest the need for doing listening testing except in the case where someone asserts that crossover distortion in high quality Class AB amps is audible, in which case the onus would be on them to establish this following well-established scientific principles. Anyone who isn't inclined to make that assertion would have no reason for doing this, i.e., the burden of proof belongs not to them. For me personally, what would be more informative would be if some very knowledgeable and capable person finds or takes measurements of several high quality Class A amps and also several high quality Class AB amps, with the distortion and noise components of these measurement keep in separate boxes for all these tests, and for this to be done at low volume level, so as to provide a meaningful answer to the question of whether there is any general advantage of one class vs. the other at low playback level specifically. It is a question that I sometimes wonder about, but I'm not someone who does or has ever done this kind of testing, so I'll never know the answer unless someone else who is more capable is equally curious about the same question.

 

[SIY]

For me personally, what would be more informative would be if some very knowledgeable and capable person finds or takes measurements of several high quality Class A amps and also several high quality Class AB amps, with the distortion and noise components of these measurement keep in separate boxes for all these tests, and for this to be done at low volume level, so as to provide a meaningful answer to the question of whether there is any general advantage of one class vs. the other at low playback level specifically.

This is done routinely in reviews. Here's an example from an AB amp- and of course, no crossover distortion (residual magnified 1000x).

 

[MrPeabody]

Or if it even exists in detectable quantities in engineered products. I don't see it in any of the low power or high power traces from any of the AB amps that have cycled through here in the past few years.

Self has some good analyses and data on this.

I haven't seen it either, but I'm not entirely certain that I've looked closely enough to warrant making a summary statement on the question. But it does seem to me that if there is any true advantage of Class A a low playback level, it would be something that is widely known. If you or anyone else looks into this question more closely, I will suggest trying to keep noise and distortion separate, because I have a fuzzy sense that this might be important at low playback level, more so than at higher playback level. For example, it may be (I'm just thinking out loud) that at low playback level, distortion is high and noise is low for AB and the reverse is true for A. I'm not suggesting that this is the actual case. I'm just trying to give an example that would maybe illustrate why it might matter more for distortion and noise to be measured and represented independently for low volume than it does for high volume. Well, to be honest I'm not at all sure this makes sense, but I think it might.

 

[MrPeabody]

This is done routinely in reviews. Here's an example from an AB amp- and of course, no crossover distortion (residual magnified 1000x).

View attachment 100350

This is interesting, but I could use some help interpreting it. If I did the math correctly, the signal corresponds roughly to 20 milliWatt (RMS or average power) for a 4 Ohm load. Certainly this would be considered a very low playback level. Since the other trace appears random I assume it is noise. Dividing, squaring and converting to dB, it appears to me that the noise is about -15 dB compared to the signal ... is this about right or is the completely wrong? I don't know. And whether this is about right or completely wrong, the important question is whether there are any typical, inherent differences between A and AB from this standpoint.

 

[SIY]

This is interesting, but I could use some help interpreting it. If I did the math correctly, the signal corresponds roughly to 20 milliWatt (RMS or average power) for a 4 Ohm load. Certainly this would be considered a very low playback level. Since the other trace appears random I assume it is noise. Dividing, squaring and converting to dB, it appears to me that the noise is about -15 dB compared to the signal ... is this about right or is the completely wrong? I don't know. And whether this is about right or completely wrong, the important question is whether there are any typical, inherent differences between A and AB from this standpoint.

See a recurring spike at the zero crossing? No? Then there's no crossover distortion.