Audiophiles, generally don't like class D amps!

[voodooless]

"basically invented the field" is a bit of a stretch.

Yeah, there were some older foundations for sure. Let’s call it “popularized the field”. It was basically since the invention of the UCD at Philips that performance levels became good enough to actually compete with the other classes.

I do understand what you're getting at (and, heck, yeah, PWM is all analog, all the time -- good old triangle waves doin' the heavy liftin'),

You can make PWM in different ways, even digitally. It will be a discrete PWM signal then with quantized time steps.

But this doesn’t make it an amplifier. All the bits after it are pure analog. And to achieve good performance you’ll need feedback, and feedback you cannot do in the digital domain. You’ll need to do a conversion somewhere.

Basically the only implementation of a mostly digital Class D amplifier with decent performance is from Axign. Even here the feedback loop has an ADC in it to function.

And this is a major exception. Basically any other digital input Class D amp chip will have a conventional DAC integrated and a fully analog Class D stage behind it.

And one example doesn’t prove that Class D is digital. It just proves that you can be more digital than others. Nothing more, nothing less. It’s like saying that all light switches are digital because someone made one with Bluetooth.

 

[Galliardist]

Digitally controlled switching amplifiers is the best description of modern Class Ds.

That may well be the right answer.

But what if you have a digitally/microprocessor controlled turntable - do we call that digital? Digital control is not the same as digital signal, whether the thing being controlled is a turntable, a class AB amp, a class D amp module...

Interestingly, Marantz have tried also to argue that in the current reference SACD players, the DSD signal that they convert to analogue is actually an analogue signal as well.

Presumably they wanted to muddy the waters...

 

[DonH56]

I've had this debate for as long as I can remember, and it never fails to hit the hornet's nest. It beggars belief how so many people who consider themselves well educated continue to insist that class D isn't digital. First of all, it's not class D because it's digital, and neither is it digital because it's class D. That's just a coincidence.

Class A / B / C / D etc concerns the biasing of the output transistors, and that's what sets all the cardinal characteristics of an amplifier - frequency, power, bandwidth, efficiency, linearity, gain, distortion, weight, dissipation, size etc. With class D, the output transistors are biased either all on, or all off - in the voltage domain.

The argument revolves around the signal that controls those transistors in the time domain, which can be analogue. But it can also be digital, and it makes little difference.

D1. A class D amplifier with analogue drive will typically have high power and high efficiency and adequate bandwidth and linearity.
D2. A class D amplifier that has digital drive will typically have high power and high efficiency and adequate bandwidth and linearity as well.

The differences between D1 and D2 are tangible but small - like the insignificant SINAD differences that we obsess about on ASR. Both D1 and D2 can achieve high bandwidth, dynamic range and linearity etc.

However, both D1 and D2 are substantially different to class A and B in all the other cardinal characteristics - size, weight, power etc. There are big differences between the amplifier classes, which are down to the nature of the transistor biasing scheme in the voltage domain, not the nature of the control signal in the time domain, which is entirely orthogonal.

It's not about generating a time varying PCM digital data stream at the transistor output like 001011101011001001010. That's digital data, which is just a subset of digital. Digital is a much simpler and more basic. It doesn't expressly mean PCM or digital data - that's different - a subset of digital. Digital just pertains to the information that is carried with a digit; finger or no finger; 0 or 1; high or low; on or off.

If the output transistor is just on or off, that is digital, and that's what gives a class D amp all of it's most important characteristics.

Yup, all those uneducated people like A.H. Reeves and all the folk at the IEEE, IRE, and so forth who defined the amplifier classes must be wrong. The pulse width depends upon the input signal, a continuous variation, so is not digitized (i.e. not digital). Amplitude is continuous after the output filter. Digital signals are quantized in time and amplitude. So to me it is fundamentally an analog amplifier. As one of the uneducated beggarly masses I'll go on believing what I've been taught the past 40+ years.

 

[mhardy6647]

Presumably they wanted to muddy the waters...

an audiophile analog favorite since (at least) the early 1960s!

 

[antcollinet]

Information can be 'carried' with the operation of a single light switch. Ever heard of morse code?

Of course it can be. But

1 - I've never heard of morse code being described as digital.
2 - In a standard light switch setting it doesn't. It simply controls a light. No digits involved.

Light switches are never described as being digital, and nor for that matter are light bulbs, despite both operating in one of two states - on or off.

 

[voodooless]

1 - I've never heard of morse code being described as digital.

I think for all intents and purposes, we can describe Morse code as digital. It's a modulation scheme for transmitting digital information, very similar to the way RS232 transmits bits.

 

[Galliardist]

I think for all intents and purposes, we can describe Morse code as digital. It's a modulation scheme for transmitting digital information, very similar to the way RS232 transmits bits.

I presume anything that can carry numeric codes can become digital.
https://www.rfc-editor.org/rfc/rfc2549 - carrier pigeons

However, Morse code uses time, not numbers. I'd suggest therefore that it is not digital. The fact that it is modulated does not make it any more digital than AM radio...

 

[Mnyb]

I presume anything that can carry numeric codes can become digital.
https://www.rfc-editor.org/rfc/rfc2549 - carrier pigeons

However, Morse code uses time, not numbers. I'd suggest therefore that it is not digital. The fact that it is modulated does not make it any more digital than AM radio...

Imho it is but the encoding decoding takes place in the morse operators head or with the help of a table

 

[Mnyb]

Back on track where the dunning kruger aflicted people insikts that class D is digital because it begins with the letter D that’s al it is .

I’ll bring a permanent marker and scribble “D” on some tube amps and LP’s and see if they come after me with pitchforks.

Digitaly controlled switching is very common where I work with VSD drives.. it’s

 

[antcollinet]

It's a modulation scheme for transmitting digital information

It's a coding scheme transmitting symbols representing characters. The information is no more digital than the letters on the page of a book.

There isn't even a numeric/digital representation of the characters like in Ascii.

By that definition semaphore with flags would also be digital. Or smoke signals.


Perhaps the issue is with the definition of digital. For me - for something to be digital there has to be numeric coding involved - not just on/off signalling.

 

[voodooless]

There isn't even a numeric/digital representation of the characters like in Ascii.

Yes, there is:

make a dot 0 and a dash 1, and you have a variable-length character encoding system.

 

[voodooless]

However, Morse code uses time, not numbers. I'd suggest therefore that it is not digital.

A lot of digital things use time... Do you think RS232 is analog?

 

[antcollinet]

A lot of digital things use time... Do you think RS232 is analog?

RS232 transmits sequences of bits (typically 8 bits plus framing) making up a binary number. For me it is the encoding of numbers and the subsequent use of those numbers to represent other stuff that makes something digital.

 

[voodooless]

RS232 transmits sequences of bits (typically 8 bits plus framing) making up a binary number. For me it is the encoding of numbers and the subsequent use of those numbers to represent other stuff that makes something digital.

I don't see the difference here.. the encoding scheme is just different. The purpose is exactly the same.

 

[antcollinet]

I don't see the difference here.. the encoding scheme is just different. The purpose is exactly the same.

Numbers or not numbers.

The morse symbol for B : Dash dot dot dot is not a number. If they were actual binary digits, it would be 8. But that is not what it means. The symbol for both E (dot) and I (dot dot) would be the same (zero). But they aren't.

The ascii symbol for B : binary 01000010 = decimal 66, and that is how it is represented in Ascii tables.

 

[voodooless]

Numbers or not numbers.

The morse symbol for B : Dash dot dot dot is not a number. If they were actual binary digits, it would be 8. But that is not what it means. The symbol for both E (dot) and I (dot dot) would be the same (zero). But they aren't.

I can assign numbers to these letters just fine.. It doesn't matter. That B is represented by the number 8 is arbitrary and irrelevant. That we use fixed 8 bits for an ASCII character (or 7) is arbitrary and irrelevant. I can encode numbers, text, or other stuff in any way I like in binary code. It's all still binary. The fact that there is intra-character additional time is also irrelevant. It's just part of the protocol, just like we have a stop bit in RS232.

 

[DanielT]

an audiophile analog favorite since (at least) the early 1960s!
View attachment 347711

Here is one of my favorite blues records, damn good I must say:

Muddy Waters Sings "Big Bill" - Wikipedia

en.wikipedia.org

Arguing about class D amps with class D hating audiophiles is not for me. Class D is not part of their belief system. Nothing they worship. You do not convince a believer with objective facts to abandon his faith. They can believe what they want, as long as they don't bother me. I stand for freedom of (amp) religion. Besides, if they are happy in their faith so, hey way not.

 

[Mnyb]

Here is one of my favorite blues records, damn good I must say:
View attachment 347757

Muddy Waters Sings "Big Bill" - Wikipedia

en.wikipedia.org

Arguing about class D amps with class D hating audiophiles is not for me. Class D is not part of their belief system. Nothing they worship. You do not convince a believer with objective facts to abandon his faith. They can believe what they want, as long as they don't bother me. I stand for freedom of (amp) religion. Besides, if they are happy in their faith so, hey way not.

I’ll wait until we have class-F amplifiers for audio and see what the religious makes up about that

 

[Galliardist]

I’ll wait until we have class-F amplifiers for audio and see what the religious makes up about that

Maybe someone could make a version of class D that uses lots of power all the time and call it Superspeed-A?

 

[Galliardist]

I can assign numbers to these letters just fine.. It doesn't matter. That B is represented by the number 8 is arbitrary and irrelevant. That we use fixed 8 bits for an ASCII character (or 7) is arbitrary and irrelevant. I can encode numbers, text, or other stuff in any way I like in binary code. It's all still binary. The fact that there is intra-character additional time is also irrelevant. It's just part of the protocol, just like we have a stop bit in RS232.

I thought we were arguing about digital, not binary. If you assign a number to something, it only works if everyone else does as well, so we can understand and translate it as a number as well as giving those numbers other meanings.

As it happens, Morse code has either four or five values, according to definition:
a long "on", dash
a short "on", dot
a long "off" between letters
a longer "off" between words
and arguably, a very short "off" at the end of a dash or dot (the very short off can be said to be part of the dash or the dot)

It's another example, going back to an earlier comment of mine, of something that is arguably digital or not, but is not in fact binary when it comes to definition and meaning. Even though the key is either down or up...

There should be some information theory that applies to this, separating the binary "on off" from the values, but I can't remember the terminology used.