"William Sommerwerck" wrote in message
...
The Web is wrong. Most switching amps are analog. That is,
everything varies continuously, rather than in quantized steps.

That is a confusing and not particularly true statement.

It might be confusing if you've been brainwashed into thinking "pulses" =
"digital", but it is nevertheless true.


By the way, Arfa, you're doing something intellectually invalid --
you're
"appealing to authority", rather than thinking for yourself, or
explaining
what's going on.

I am not.

Then why did you post Web references as examples of what other people
think?
Majority opinion is proof of nothing.


Because on average, in the real world, if "millions of people" believe
something, and one does not, it is not the millions who actually *are*
wrong.

On average. But there are exceptions. The world is not 6000 years old. Yet
millions of people believe that. The majority is not the "authority".


Oh puh - leees ...




I'm not really quite sure exactly what you're saying here about the class
D
amplifier. The analogue input signal is not converted directly to any
kind
of 'value' represented by a binary number, such as might be the case if
you ran it through a traditional A-D converter. Instead, it is run
through
a
comparator, with a triangle wave as the reference input. This results in
direct conversion to a PWM signal. I accept that this does not represent
'quantization' as such, so is not producing a "truly digital" signal, but
I
also do not believe that once the signal is in PWM form, it can either be
considered to be analogue any more.

Ah! Here's the problem. It's the confusion between /waveform/ and /data/.

A pulse is just a pulse. In and of itself it means nothing. It is neither
"digital" nor "analog" -- it's just a waveform.

The issue here is how we modify a waveform to transmit data.

Suppose we sampled a signal at or above the Nyquist rate and transmitted
each sampled value as a pulse of that value. (This is easily done with a
sample-and-hold circuit.)

How is the /data/ in that series of pulses represented? Well, it varies
/continuously/, just as the original signal did. It has not been
quantized,
so it cannot be represented as one of a /finite/ group of numbers. That's
analog -- continuous variation.

OK. I understand why you might contend that a PWM signal is an alternative
analogue version of the original (conventionally understood) analogue
signal. However, I still believe that calling a class D amplifier
"analogue", and insisting that it is not in any way 'digital' is likely to
be confusing to the vast majority of conventionally schooled electronic
service engineers, as opposed to those who have sufficient understanding and
interest in the math of signal digitization and processing to feel
otherwise.

Rightly or wrongly, most service engineers understand an analogue signal as
what you would conventionally see on a 'scope, if you put it across an
amplifier's speaker terminals, whereas a signal that varies between two
levels only, irrespective of how the pulse width is varying, tends to be
considered as digital, due to the 'conventional' understanding that simple
service engineers have, of the operation of logic circuits (as for ECL, that
was a special case that most will never have heard of anyway, and as I
recall, the 'pulses' were nothing like a sine wave, and actually difficult
to distinguish from the noise floor).

As far as my quoting web references goes, most normal people consider this
resource to be the repository of all human knowledge, and the dog's ********
of reference media. Whilst it is of course not always right on everything,
where there is collected opinion from many many different and respected
sources, and that opinion is broadly consistent, surely any reasonable
person could not be considered stupid, or without thought of their own, for
accepting it as a lesson. How else do we learn about any subject other than
to either research it, or be taught it by someone considerd to be an expert
? Thus, I make no apology for using the 'net as a research tool, and for
citing links to the data I have found.

I still contend that there is no real name for the process employed in a
class D amplifier. I don't believe that it is analogue in the conventionally
understood sense (and I really don't care if you and bz feel that makes me
"mistook" was it he called me ?) and if you want to be purist about the math
of quantization, neither is it digital in the true sense.

Perhaps we need to coin a new name for it. As it's similar in concept to a
switch mode power supply, maybe we should call it a 'switch mode amplifier'.

And that really is all the time that I want to waste on this. I know what I
mean, and I rather think that most conventional engineers on here do too,
and understand quite well what is implied when calling a class D amplifier,
digital ...

Arfa

snip rest