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

[dualazmak]

... and 120Hz

Here in east part of Japan, electricity is 100 V 50 Hz, and in west part of Japan it is 100 V 60 Hz.

It is interesting that (as far as I know) only microwave ovens do not compatible for both of 50 Hz and 60 Hz. I moved west to east, east to west, and then again west to east, three times, in past 30 years, and every time we needed to purchase a new microwave oven for our kitchen!

 

[pma]

Here in east part of Japan, electricity is 100 V 50 Hz, and in west part of Japan it is 100 V 60 Hz.

I know about it. I have installed several plasma spray systems in Japan. Kitakyushu area, Osaka area and Nagoya area. All of them have 60 Hz. Here in central Europe we have 50 Hz.

 

[GXAlan]

The PM-90 is basically a PM-88se built on the PM-95's die-cast chassis with a fold down flap instead of visible front panel controls. It was less than half the price of the PM-95 and lacks a huge amount of cost that went into that unit. The PM-95 has (my count in my unit when I restored it) 27 gas-filled relays just for source selection and signal routing. The D/A converter section is the same as the CD12/DA12LE with a hand selected TDA-1541S1- they took the S1 chips and picked the very best ones for the DA-12 and the PM-95 because Bitstream was where Philips was at..

Time to get the PM-95 to Amir for testing then!

They all look very different with pricing very inconsistent. The PM-99SE which was a HDAM version of the PM-90 was 330,000 yen compared to the 160,000 yen of the PM-90 or 318,000 yen of your PM-95. The PM-88se looks very different.

PM-88se


PM-90


PM-95


PM-88SE / PM-90 / PM-95 / PM-99SE
THD. 0.008%. / 0.0015% / 0.008 % / 0.005%
SNR. 107 dB. / 111 dB / 106 dB / 111 dB

Either way, it will be interesting to see Class A vs AB under an APx555 at 5W.

 

[MakeMineVinyl]

Do you remember cassette decks back in the day had arced and pre-drilled threaded holes in the chassis for rotating/shifting the transformers individually to ensure the lowest levels of hum and buzz from transformers? The good old days.

Reel to reel decks like my Sony 660 had Mu metal shields which were brought over the face of the playback head to reduce hum. These could be adjusted and rotated to get the minimum amount of hum. More upscale recorders fully enclosed the playback head. You already know this, but I'm just throwing it out for those who don't.

One of the early Ampex models placed the capstan motor on the 'wrong' side so that it was as far away from the playback head as possible. It not only looked weird, but it was found to be unnecessary.

 

[pma]

The Cambridge's distortion was below the AP's residual. The Parasound wasn't quite as good, though still reasonably decent.
View attachment 153148

Sometimes the orientation of the power transformer is slightly off, and this can cause measurable increases in AC related noise. If the transformer is a toroid, rotating it to find a null can help.

Yes, exactly what you expect from radiation pickup. If it had been my amp, I'd be rotating the transformer to see if I could get a null. But that's a step too far for a review loaner.

I am still almost sure that the problem is not a transformer radiation, but poor PCB design. Especially ground returns (signal and bypass electrolytes) and possibly supply rails induction. As a proof, I am attaching a quick measurement of one of my amplifier samples that is supplied from a 270VA transformer and filter/rectifier that are in 2m distance from the amp so the radiated field (from the transformer) cannot affect the amp board. Of course PSU ripple voltage and half-wave current pulses are still there on the PCB.

Two PCB ground return schemes, #1 and #2 have been measured, none of them is ideal but the difference is stunning, and #1 resembles the JC5 result. John should hire a specialist in power amp PCB design .

Grounding scheme #1

Grounding scheme #2

 

[SIY]

I am still almost sure that the problem is not a transformer radiation, but poor PCB design. Especially ground returns (signal and bypass electrolytes) and possibly supply rails induction. As a proof, I am attaching a quick measurement of one of my amplifier samples that is supplied from a 270VA transformer and filter/rectifier that are in 2m distance from the amp so the radiated field cannot affect the amp board.

Two PCB ground return schemes, #1 and #2 have been measured, none of them is ideal but the difference is stunning, and #1 resembles the JC5 result. John should hire a specialist in power amp PCB design .

Grounding scheme #1
View attachment 153467

Grounding scheme #2
View attachment 153468

Since you didn't have this amp to measure, it's complete speculation rather than "a proof." It could be one, could be the other, could be some combination of both.

Last time I pointed out a physical issue with one of his amps (one I had recommended to my at-the-time boss), John sneered that at $2000-3000, this was mid-fi and people shouldn't complain.

 

[pma]

Since you didn't have this amp to measure, it's complete speculation rather than "a proof." It could be one, could be the other, could be some combination of both.

It is based on numerous amps measured and numerous amps designed. As a circuit designer, I can quite easily tell from the measurements seen where is the problem. That's a result of experience and active engineering (electrical, not chemical). That's a difference between a designer and a reviewer. Reviewer can only speculate, unless he has not his own design and test experience in the field. It can often be seen in ASR reviews, measurements are OK but the explanations are wrong. Such is life, none of us knows everything, neither you nor me .

 

[SIY]

It is based on numerous amps measured and numerous amps designed. As a circuit designer, I can quite easily tell from the measurements seen where is the problem. That's a result of experience and active engineering (electrical, not chemical). That's a difference between a designer and a reviewer. Reviewer can only speculate, unless he has not his own design and test experience in the field. It can often be seen in ASR reviews, measurements are OK but the explanations are wrong. Such is life, none of us knows everything, neither you nor me .

Yes, that must be it, I haven't designed anything.

Sheesh, Pavel.

 

[pma]

My replies related to amplifier output hum components are in a new thread

https://www.audiosciencereview.com/...ated-hum-and-noise-in-power-amplifiers.26645/

 

[Trdat]

need to disconnect your preamp from the poweramp.
use a piece of wire to connect the centre pin of phono input to the outer surround or a part of metal casing.
do the same with XLR inputs as he said too.
turn on the amp, and see if hum persists.

Can someone confirm this instruction. I have a class AB amp and I also have a buzz through my compression drivers. After trying and meddling around with a few things I am guessing its the amp. Using a Hypex amp there is no buzz at all. My curent amp is the NAD356 and unfortunately has no XLR option.

So can I take any RCA input say Aux or CD and then connect it to the metal casing?

Am I using a RCA wire into the input and getting the other end of the wire to touch the metal casing? And can I use one channel of the inputs?

 

[horias2000]

You don't really have to choose one or the other . I've built one power amplifier that can work in both class AB/B and class A. It is based on the Trimodal Amplifier by Douglas Self. I've performed THD measurements and you can clearly see the odd harmonics go down when switching to class A. In general, a well designed class AB/B power amplifier can deliver very low THD and you will not hear the difference. By the way, using the amplifier I mentioned above and switching between classes, I could not hear any difference. I will try to add THD measurements but unfortunately I do not have a good toolchain to measure THD at the moment. I use a Scarlet Solo USB ADC/DAC and REW. I do not have enough repeatability as each time I get different values and this bothers me. Once I figure out what's going on I will post THD measurements with the two classes. BTW @pma , whst interface are you using to measure THD?

 

[pma]

I would like to show, in measurement examples, what makes the biggest difference between low-distortion class A and class AB. Measurements are made on a class A amplifier module with idle current Iq = 1.4A and 4ohm load. With this idle current and load, and single sine wave input, amplifier works up to 15.68W/4ohm in class A and above 15.68W it transits into class AB. Below 15.68W there is no crossover and switching distortion and distortion spectrum has only the lowest order harmonics, H2 and H3. In fact, in my setup we shall only see the residual distortion of the soundcard and noise due to REW Java operation in 16bits. Unfortunately, this soundcard does not work with ASIO in REW, however is free of usual high order distortion components that would mess the subject.

Above 15.68W of the amp power, crossover and switching distortion appears and we cab see high order harmonics in the spectrum. Because of ear masking curves, high order harmonics are much more audible than low order harmonics. However, here they stay well below audibility threshold.

class A operation

transition into class AB

class AB operation (this is similar what we measure with TPA3255 class D amplifiers)


Intermodulation twin-tone measurement is even better indicator of the transition into class AB. We need to keep in mind that the twin-tone with the same amplitude as a single-sine wave has only half power of the sine wave.

class A operation

close to class A limit

class AB operation

Deeper in class AB we can see "skirts" of the odd harmonics near to original two tones. This is a clear indication of crossover and switching distortion. The same spectrum we can see with TPA3255 based class D amplifiers. There is a direct line between class D high frequency nonlinearity (due to most probably dead zone) and class AB crossover/switching distortion.
Class A has the purest spectrum of all amplifier operating classes with the advantage of excellent HF linearity. This of course applies only for a careful class A circuit designs and not for the simplest circuits like ZEN, F1 and similar.

 

[Ken Tajalli]

People are gonna come at you with measurements of numerous class AB amplifiers beating the results above.
However, on a like for like test, your results are valid.
Like for like means, every ingenuity aforded to a class AB to make it measure better, be afforded to a class A at wattages comfortable for both.
After all, there is no denying class AB amps are more efficient.
It is no good comparing an amp of my design and build, to, say, one of a major manufacturer.

 

[DonH56]

@pma How do you know the higher odd-order distortion terms are explicitly from crossover distortion and not just normal increasing nonlinearity at higher power? When I did something like this in the primordial past, I kept the power output the same, and adjusted the output stage bias. When crossover distortion appeared, it added higher harmonics and some HF noise stuff, but did not seem to affect the primary distortion all that much, although second-order terms rose since crossover glitches appeared at every zero crossing. I had a hard time isolating crossover distortion in the FFT though it was clear on a 'scope.

Curious - Don

 

[tomchr]

@pma Is this for an amp that's optimally biased for Class AB or one with a high bias to allow for a larger Class A region?

Tom

 

[NiagaraPete]

Amps are like maple syrup, class or grade don't correspond to goodness.

What about Canadian maple syrup Vs US maple syrup. I'd have to argue for Canadian.
<my 2 cents>A Vs AB. Very few companies even look at class A design anymore. One that does is Luxman and they have some stuff I'd love but can't afford.</my 2 cents>

 

[Raindog123]

Very few companies even look at class A design anymore

What about multi-stage designs? My integrated NAD C375BEE (Canadian, btw) is a class A all but the final power stage (that is AB)..?

 

[mocenigo]

There is a direct line between class D high frequency nonlinearity (due to most probably dead zone) and class AB crossover/switching distortion.

I do not see how. They are completely different and uncorrelated phenomena. So the TPA3255 based Class D amps have a similar pattern of intermodulation distortion as the AB you measured. This is not representative of all Class D amps. Also I am not sure it is a self-oscillating design, that usually have better distortion patterns. And it would be interesting to know at which power. TI claims it can produce 600W but at 10% distortion...

Class A has the purest spectrum of all amplifier operating classes with the advantage of excellent HF linearity. This of course applies only for a careful class A circuit designs and not for the simplest circuits like ZEN, F1 and similar.

*good* class A can measure as well as any other topology, but at higher power and material cost per output Watt. Your measurements are great though.

 

[Ken Tajalli]

Is there such a thing as class AB interstage , buffer stage ?

 

[mocenigo]

Is there such a thing as class AB interstage , buffer stage ?

Rare, but yes. Some opamps work in AB close to their maximum output.