Have amplifiers finished evolving?

[Arkless Electronics]

Class D definitely not digital no.

Just repeating my previous post from a couple of days ago in case some thought it a discussion point rather than a statement of fact.

 

[Arkless Electronics]

Jez,
The class-A-versus-class-A/B and feedback things. Are you saying that class A is better because it needs less and/or different feedback? Thanks.

Indeed yes it requires much less feedback as it is more linear in the first place. But it's only one of many areas in which class A is better. In fact. as I said earlier, class A is technically superior in every way to all other classes.

 

[Shadders]

Class D definitely not digital no.

Just repeating my previous post from a couple of days ago in case some thought it a discussion point rather than a statement of fact.

Hi,
I am interested to know why you state class D is not a digital amplifier. Can you reference the relevant mathematical theorems indicating that the output signal is not digital ?. Thanks.

Regards,
Shadders.

 

[Arkless Electronics]

Hi,
I am interested to know why you state class D is not a digital amplifier. Can you reference the relevant mathematical theorems indicating that the output signal is not digital ?. Thanks.

Regards,
Shadders.

Too obvious to be worth discussing. So I won't.

 

[Shadders]

Too obvious to be worth discussing. So I won't.

Hi,
Sorry, but i am not sure how you have come to the conclusion as why it is obvious.

The output signal or switching signal from the MOSFET switching pair node is a binary digital signal. It meets the mathematical criteria for being a digital signal, and would be classified as a binary digital signal. Please see the following wiki links :

https://en.wikipedia.org/wiki/Digital_geometry
https://en.wikipedia.org/wiki/Discrete_space
https://en.wikipedia.org/wiki/Digital_signal

The above links provide all the mathematical proof that the signal is a binary digital signal.

As such, i am interested to know your mathematical proof that the output signal is not digital, since it is not obvious. Can you provide a wiki link to support your theory ?. Thanks.

Regards,
Shadders.

 

[Darth Vader]

Hi,
I am interested to know why you state class D is not a digital amplifier. Can you reference the relevant mathematical theorems indicating that the output signal is not digital ?. Thanks.

Regards,
Shadders.

Its not digital. The analogue input modulates a 'carrier' frequency and in this case the modulation only alters the width of the pulses. There is no conversion of the analogue signals into binary data. That can be done to carry data and then it becomes PCM but thats not how a class D amp works.

Cheers,

DV

 

[Shadders]

Its not digital. The analogue input modulates a 'carrier' frequency and in this case the modulation only alters the width of the pulses. There is no conversion of the analogue signals into binary data. That can be done to carry data and then it becomes PCM but thats not how a class D amp works.

Cheers,

DV

Hi,
Thanks.

The fact that the PWM is an asynchronous PWM, does not stop the output signal being classified as a digital signal. There are many asynchronous digital circuits. They are never classed as analogue because they are asynchronous.

From the wiki page :

https://en.wikipedia.org/wiki/Pulse-width_modulation

"Pulse-width modulation (PWM), or pulse-duration modulation (PDM), is a fancy way of describing a digital (binary/discrete) signal that was created through a modulation technique, which involves encoding a message into a pulsing signal."

It clearly states it is "a fancy way of describing a digital (binary/discrete) signal". So even the wiki page describes PWM as a binary/discrete signal - which meets the criteria for being a digital signal.

A digital signal is defined by the permitted levels, which are discrete, and as per the wiki links above in another post (discrete space, digital_geometry, digital_signal), the output of the class D amplifier is indeed, digital.

Regards,
Shadders.

 

[Darth Vader]

So you really have no idea and that is a poor description. Don't believe everything on the web is accurate or in some cases even true.

Cheers,

DV

 

[Jim Audiomisc]

This is also the same for the class D amplifier. Regardless of the signal behaviour, the signal is still classified as digital, as it can only be one of two values.


The continuously variable you refer to is the analogue input to the comparator of the amplifier which is part of the PWM modulator, which forms part of the control logic of the output switching. Whether this PWM modulator is analogue or discrete, asynchronous or synchronous, is irrelevant since the output stage is a binary digital signal.

Again, people using 'digital' as if it were a synonym for 'binary'. It isn't.

And as has been pointed out, if the system operates on the basis of contunuously variable pulse widths, it isn't 'digital' because digital systems have to be quantised in the quantity the use. In this case continuous 'width' variations.

 

[Shadders]

So you really have no idea and that is a poor description. Don't believe everything on the web is accurate or in some cases even true.

Cheers,

DV

Hi,

Will you be updating that page to correct the serious errors ?

Do you have a reference which you deem more accurate ?. Would a communications book be acceptable ?

Edit :
I have the following books on communication theory - Communications Systems (Carlson), Modern Analogue and Digital Communications systems (Lathi), Principles of Communications (Ziemer/Tranter), and all describe Pulse Width Modulation which is also know as Pulse Duration Modulation, as a digital signal.

The signal meets the mathematical criteria for a binary digital signal.

I really would like to see the proof of why the signal at the transistor switching node is not a binary digital signal. Thanks.

Regards,
Shadders.

 

[Jim Audiomisc]

Aside than instability am I correct in thinking the Class D amps are setup for a chosen load ie typically 4R or 6R? Into other loads I believe there will be changes to the high frequency response. This is believe causes me issues with my 16R speakers.

It is the case that the output network will have an effect that varies with the loudspeaker impedance. So if no form of overally feedback or correction for that is employed then, yes you can expect the HF response, etc, to vary with the choice of speaker.

 

[Jim Audiomisc]

Hi,
Sorry, but i am not sure how you have come to the conclusion as why it is obvious.

The output signal or switching signal from the MOSFET switching pair node is a binary digital signal. It meets the mathematical criteria for being a digital signal, and would be classified as a binary digital signal. Please see the following wiki links :

https://en.wikipedia.org/wiki/Digital_geometry
https://en.wikipedia.org/wiki/Discrete_space
https://en.wikipedia.org/wiki/Digital_signal

The above links provide all the mathematical proof that the signal is a binary digital signal.
.

Actually it shows that it can be unwise to use or misunderstand what either Google or Wikipedia say.

Binary simply means the use of two discrete levels or quantities.

So an amp that uses two states/quantities and flips between them may be binary.

Digital doesn't specify the number of such quantised states used for representation.

So if an amp uses, say, two voltage levels *but can continuosly vary their durations* then it isn't 'digital' even though the choice of *levels* is binary. The reason being this uses continuous width adjustments (not quantised ones) to represent the output information.

 

[Jim Audiomisc]

Hi,
Thanks.

The fact that the PWM is an asynchronous PWM, does not stop the output signal being classified as a digital signal. There are many asynchronous digital circuits. They are never classed as analogue because they are asynchronous.

From the wiki page :

https://en.wikipedia.org/wiki/Pulse-width_modulation

"Pulse-width modulation (PWM), or pulse-duration modulation (PDM), is a fancy way of describing a digital (binary/discrete) signal that was created through a modulation technique, which involves encoding a message into a pulsing signal."

It clearly states it is "a fancy way of describing a digital (binary/discrete) signal". So even the wiki page describes PWM as a binary/discrete signal - which meets the criteria for being a digital signal.

A digital signal is defined by the permitted levels, which are discrete, and as per the wiki links above in another post (discrete space, digital_geometry, digital_signal), the output of the class D amplifier is indeed, digital.

Regards,
Shadders.

It is a shame that people have come to take for granted that whatever appears on wikpedia must be correct, or simply mis-apply it. But here the key point is that this depends on if the *durations* are quantised or not because the durations are being varied to convey the information.

 

[Jim Audiomisc]

Hi,

Will you be updating that page to correct the serious errors ?

Do you have a reference which you deem more accurate ?. Would a communications book be acceptable ?

(snip book list)

I really would like to see the proof of why the signal at the transistor switching node is not a binary digital signal. Thanks.

FWIW personally I decided years ago that trying to correct wikipedia errors or wordings that mislead would be a lifetime quest, so I chickened out on that.

I was going to suggest that you could at least start with

http://jcgl.orpheusweb.co.uk/InformationAndMeasurement_PDF_Book_pf.pdf

but I see you've already run into the common phenomenon where textbooks also propagate things even when it may mislead.

I can't comment on the specific design you and others have referred to as I don't know the details. But the basic points I've made about the use of the terms are correct. That this is misunderstood doesn't surprise me given that so many engineers seem unware of the advantage 3-state systems have over binary when it comes to generating bipolar analog output waveforms.

 

[Shadders]

Actually it shows that it can be unwise to use or misunderstand what either Google or Wikipedia say.

Binary simply means the use of two discrete levels or quantities.

So an amp that uses two states/quantities and flips between them may be binary.

Digital doesn't specify the number of such quantised states used for representation.

So if an amp uses, say, two voltage levels *but can continuosly vary their durations* then it isn't 'digital' even though the choice of *levels* is binary. The reason being this uses continuous width adjustments (not quantised ones) to represent the output information.

Hi,
Yes - i did read that aspect - that the value being transferred has to be discrete based on the information for it to be digital. But then it also defines the meaning of an analogue signal - and the signal at the output of the transistor switching node is certainly not analogue. So, a conundrum.

Wiki does provide the basis for discussion - so everyone has the same reference with which to discuss.

Regards,
Shadders.

 

[Shadders]

FWIW personally I decided years ago that trying to correct wikipedia errors or wordings that mislead would be a lifetime quest, so I chickened out on that.

I was going to suggest that you could at least start with

http://jcgl.orpheusweb.co.uk/InformationAndMeasurement_PDF_Book_pf.pdf

but I see you've already run into the common phenomenon where textbooks also propagate things even when it may mislead.

I can't comment on the specific design you and others have referred to as I don't know the details. But the basic points I've made about the use of the terms are correct. That this is misunderstood doesn't surprise me given that so many engineers seem unware of the advantage 3-state systems have over binary when it comes to generating bipolar analog output waveforms.

Hi,
In one way, yes, the textbooks may repeat the same misleading approach. The amplifier output is certainly NOT an analogue continuous signal - it has two discontinuities every pulse time-base duration (period).

The output signal of the switching node is constrained by the switching pulse time-base width (period). It is creating a discrete time based signal which is an asynchronous pulse width. So although the output is derived from an analogue input signal, using an asynchronous PWM, it is not an analogue signal, since the PWM has converted the analogue signal into a discrete time based binary signal.

The issue i suppose is the term digital, and if digital is strictly where logic levels represent discrete values, then the output does not meet this definition of digital. The converse, is that the signal in not analogue - as analogue is defined as continuous in time, and the amplifier output has discontinuities. Therefore there is no analogue class D.

Regards,
Shadders.

 

[JensenHealey]

You should not get hung up on the signal at the output of the switching transistor. That is not the end of work - filtration is required. Usually done explicitly with a filter network or implicitly by using the cable and speaker to do the work. To regard otherwise is to assume that the filter or speaker has an, effectively, infinitely wide bandwidth.

Switching power supplies are the same - they are not 'digital' either. Jim is right - the system could be well described as binary.

 

[Darth Vader]

Just because you can have two states on or off doesn't in itself mean anything and can't be assumed to be digital. True, on and off can be represented by 0 and 1. In Boolean arithmetic these relate to true or false and this is logic not maths. In fact in computing we often refer to OR, AND and XOR as 'logicals'. In a class D amplifier the output transistors are just acting like a fast on/off switch. The reason for this is to reduce power dissipation in the output pair so that they stay firmly in their SOA. It also means that you can get away with smaller components and less heat sinking if any. I seem to recollect that my Sinclair Class D didn't have any heat sink.

Also there are different ways in which a class D switching amplifier can be built. For example do we use a separate stable oscillator or a self oscillating stage? They both result in a pulsed output but they are different.

A class D amplifier can be simplified to input>modulator>power switching transistors>demodulator>speaker output. Nowhere in that chain is the input signal digitalised.

Thinking about it this is more closely related to traditional am/fm radio broadcasts than computers.

Cheers,

DV

 

[adamdea]

Again, people using 'digital' as if it were a synonym for 'binary'. It isn't.

And as has been pointed out, if the system operates on the basis of contunuously variable pulse widths, it isn't 'digital' because digital systems have to be quantised in the quantity the use. In this case continuous 'width' variations.

Forgive the my sophomoric ramblings but I always have to stop and think about this class D problem. And this is how I think through the irrelevance (to this question) of the binary voltage on/voltage off in class d.

If you write a line on a piece of paper then the information is encoded in a medium containing two states- ink on or ink off. But the line can either be digital, if the information represented is regarded as a number 1, or analogue if it is encoded in the length of the line. The fact that there are only two relevant states (ink on and ink off) is irrelevant to the digital/analogue nature of the information. It is a property of the medium not of the encoded information.

 

[Shadders]

You should not get hung up on the signal at the output of the switching transistor. That is not the end of work - filtration is required. Usually done explicitly with a filter network or implicitly by using the cable and speaker to do the work. To regard otherwise is to assume that the filter or speaker has an, effectively, infinitely wide bandwidth.

Switching power supplies are the same - they are not 'digital' either. Jim is right - the system could be well described as binary.

Hi,
I agree that digital is an inaccurate term, but the amplifier is certainly not an analogue Class D. No class D could ever be described as analogue if it has solely at its output a pair of switching transistors.
Regards,
Shadders.