Questions about equivalents of audio/video and digital/analog.

[Arny Krueger]

Jerry Avins wrote:

(snip)


This sounds like what I previously tried to describe as
quantized but not sampled. The signal has two states,
but the transition can happen at any time.

The signal might be thought of as being quantized in the aplitude domain,
but it is clearly not quantized in the time domain. For a signal to be
quantized, it has to be fully quantized, that is quantized in both the time
domain and the amplitude domain.

 

[Eric Jacobsen]

The voltage amplitude has nothing to do with whether
the signal is digital or analog. It can be anything,
with any characteristics you'd like to imagine.
That is because it carries no information.

So why have you kept harping on the idea of discrete voltages?

First, it is not. It varies between two voltages, and
does so continuously (and apparently too quickly for a
slow person to follow, eh?).

Can you describe a real-world two-state, "digital" signal by your
definition that doesn't behave that way?

But since the variations
contain no information and therefore do not represent
symbols of any kind, the amplitude does not determine
whether the signal is digital or analog.

You just keep drawing your circle of constraints smaller and smaller.
Yes, it is possible to keep reducing a circle of scope to vanishing
small radius, but it stops being relevant to anyone else long before
that. I'm guessing the folks left in this thread are here mostly for
amusement or with the hope of folks other than you learning something.
It's clear you're not going to get anything out of it.

We can't tell Jerry. You have not stated anything that
describes the symbols set. The information is carried
by some other characteristic of that signal (e.g., phase
or frequency). Not knowing if it carries only discrete
values from a finite set, or if the symbols are
continuously variable, we just don't know what it is.

This is *very* basic...

It's very basic obfuscation on your part. If you can't have a
meaningful dialogue with anyone without continually updating the
constraints and definitions to fit your view, you're not going to have
a lot of influence.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

 

[Jim Kelley]

Actually, they are good definitions, but they are *not*
"valid standard definitions" for this discussion.

They are valid for any discussion in which I care to use them. But
thanks for your input.

jk

 

[Floyd L. Davidson]

You do well to disagree with it. It is false. The errors that are introduced
are aliasing, not quantization errors. You could change the size of the
quantization steps any which way you want, and the aliasing would still be
there.

That is another technically *absurd* statement. Can you
explain how one gets aliasing under the circumstances
stated?

Agreed. It's an incorrect statement for the reason I stated above.

The reason you state is incorrect, and that should be
obvious.

Incidentally, I took a look at "Telecommunications
System Engineering, Third Edition" by Roger L. Freeman
to see what definitions he uses. He cited another
reference, and quoted this as "the Nyquist sampling
theorem",

"If a band-limited signal is sampled at regular
intervals of time at a rate equal to or higher than
twice the highest significant signal frequency, then
the sample contains all the information of the
original signal. The original signal may then be
reconstructed by use of a low-pass filter."

His source is "/Reference/ /Data/ /for/ /Radio/ /Engineers/,
6th Ed., ITT/Howard W. Sams, Indianapolis, 1976.

That's two more absolutely credible sources who say you need
to learn more about this.

 

[Floyd L. Davidson]

They are valid for any discussion in which I care to use
them. But thanks for your input.

That is true. You can use any definition for any word
you like, in Alice's Wonderland.

And you won't be understood by anyone else, which seems to
be the point of many who post this sort of drivel.

 

[Jim Kelley]

And you won't be understood by anyone else, which seems to
be the point of many who post this sort of drivel.

Perhaps true insofar as the word might not be understood by someone
who also does not know how to use a dictionary. Admitedly, I had not
accounted for that possibility. 8-|

jk

 

[Floyd L. Davidson]

Perhaps true insofar as the word might not be understood
by someone who also does not know how to use a
dictionary.

You, for example.

Admitedly, I had not accounted for that
possibility. 8-|

That was fairly obvious, and still is.

You are implying that any dictionary definition (hence
not your unique Alice in Wonderland definition) is
correct in any context. That is not the way a
dictionary is properly used.

"Digital", for example, has at least three different
definitions. You want to be able to pull any one of
them out of a hat, and say that it means what it
means... But that is back to Alice in Wonderland.

The word "Digital" is a Term of Art, and your common
language dictionary definition is not valid in a
technical discussion.

 

[Arny Krueger]

That is another technically *absurd* statement. Can you
explain how one gets aliasing under the circumstances
stated?

Real world signals have finite, not zero bandwidth. That means that they
have sidebands. As soon as any of the sidebands are at or exceed the nyquist
rate, there is distortion of the wave because of aliasing. Therefore the
sample rate has to be somewhat higher than the sample rate of the signal,
which is characterized by its carrier frequency.

The reason you state is incorrect, and that should be
obvious.

Incidentally, I took a look at "Telecommunications
System Engineering, Third Edition" by Roger L. Freeman
to see what definitions he uses. He cited another
reference, and quoted this as "the Nyquist sampling theorem",

"If a band-limited signal is sampled at regular
intervals of time at a rate equal to or higher than
twice the highest significant signal frequency, then
the sample contains all the information of the
original signal. The original signal may then be
reconstructed by use of a low-pass filter."

Congrats to Freeman for avoiding making a different statement that avoids
the error made in the previous statement.

 

[Jim Kelley]

You are implying that any dictionary definition (hence
not your unique Alice in Wonderland definition) is
correct in any context.

I'm not implying that at all. I'm simply saying that the terms and
definitions I used are correct in the context in which I used them.

jk

 

[Jim Kelley]

Arny Krueger wrote:

Congrats to Freeman for avoiding making a different statement that avoids
the error made in the previous statement.

Although, unless the original signal is a squarewave, saying "the
sample contains all the information of the original signal" may be
overly optimistic.

jk

 

[Floyd L. Davidson]

I've decided that it's not fruitful to continue this discussion since
the knowledge I work with admits anough understanding to get a lot of
real work done. These sorts of discussions take too much time and
produce little or no fruit.

My ability to do work does not depend on others' judgement of the
correctness of my definitions.

So you figure that posting invalid definitions to Usenet
or proving that you cannot understand standard
definitions won't be a problem for you?

It's amazing though, just who finds what with web
searches. Anyone who reads what you've had to say
here... well it could easily affect your work!

Whatever, an article with out of hand invalid statements
quoted from the two people who clearly post from an IEEE
host is a really good place to throw out something that
I've asked them for repeatedly, and they have weaseled
around the question in odd ways: IEEE definitions of
"digital" and "analog". I think it was Randy Yates who
claimed they had been posted but did admit that nobody
had ever cited IEEE as a source.

Well, it appears that the person who posted it was me.
IEEE apparently uses the standard definitions which I
have posted from other sources.

However, here is a very interesting discussion from an
IEEE dictionary:

An analog signal implies /continuity/,
as contrasted to a digital signal that
is concerned with /discrete/ states.
Often the means of carrying information
is the distinguishing feature between
analog and digital. The information
content of an analog signal is conveyed
by the value or magnitude of some
characteristics of the signal such as
phase, amplitude, frequency of the
voltage, the amplitude or duration of a
pulse, and so on. To extract the
information, it is necessary to compare
the value or magnitude of the signal to
a standard. The information content of
a digital signal is concerned with
discrete states of the signal, such as
the presence or absence of a voltage, a
contrast in the open or closed position,
or a hole or no hole in certain
positions on a card. The digital signal
is given meaning by assigning numerical
values or other information to the
various possible combinations of the
discrete states of the signal."

"The New IEEE Standard Dictionary of Electrical and
Electronic Terms", 5th ed., IEEE Std. 100-1992,
IEEE Press, New York, 1992.

I'm quoting it from Roger L. Freeman's "Telecommunications
System Engineering", 3rd ed., 1996.

For one, it clearly shows the FM-signal-through-a-limiter
example given by Jerry very clearly to be exactly as my
analysis indicated, and not at all what Jerry said.
Also they clearly state that the values assigned to a
digital symbol need not be "numerical" as someone argued
repeatedly in earlier posts.

Another example of credible references that support each
and every point that I've made. And it again highlights
that none of those saying it isn't so can find *anything*
credible to support their statements. (And that of course
is why it is not "fruitful" to argue with me. I don't buy
rotten fruit.)

 

[Floyd L. Davidson]

I'm not implying that at all. I'm simply saying that
the terms and definitions I used are correct in the
context in which I used them.

The were easily demonstrated as *not* correct in the
context used, and your whole discussion since has been
to argue the implication as stated above.

Do you read what you write?

Lets review it:

Here are some valid standard defintions:

"quantize - to subdivide into small but measurable increments."
(Merriam Webster's Collegiate Dictionary, Tenth Edition)

Your source is for common English, it is not defining
the Terms of Art used in technical discussions. This
*is* a discussion about those terms of art, not about
common English word usages. To whatever degree your
cited definition differs from the standard definitions
I've cited, you are wrong because of the context.

Here is the standardized definition for "quantization"

quantization:
A process in which the continuous range of values
of an analog signal is sampled and divided into
nonoverlapping (but not necessarily equal)
subranges, and a discrete, unique value is
assigned to each subrange.

And you looked at a invalid definition for this context and
stated:

Note that in the definition, there appears no
mention of assigning a value. Assigning a value
would then be considered a part of a separate and
distinct process of converting to digital form, as
in

Obviously the Term of Art used in technical discussion
does indeed mean there *must* be a value assigned. In
fact it makes no sense at all unless that step is
included.

Then you go on to produce other, equally invalid in this
context, definitions for other terms of art:

"digital - of, or relating to data in the form of
numerical digits",

and as opposed to

"analog - of, relating to, or being a mechanism
in which data is represented by
continuously variable physical
quantities."

The first definition states that the form must be "of
numerical digits", and that is simply unnecessary. It
must be assigned a "value". The value can be numerical
digits, but it can be otherwise too. The point is that
it must be from a finite set of discrete values. (I
used the example of flags, where the value might be a
square flag, a round flag, or a triangular flag. No
numerical digits at all.)

As you can see, knowing how to use a dictionary is
vitally important...

 

[Jim Kelley]

The were easily demonstrated as *not* correct in the
context used, and your whole discussion since has been
to argue the implication as stated above.

Do you read what you write?

Lets review it:

Here are some valid standard defintions:

"quantize - to subdivide into small but measurable increments."
(Merriam Webster's Collegiate Dictionary, Tenth Edition)

Your source is for common English, it is not defining
the Terms of Art used in technical discussions.

It perfectly defines the term as I use it.

This
*is* a discussion about those terms of art, not about
common English word usages. To whatever degree your
cited definition differs from the standard definitions
I've cited, you are wrong because of the context.

I don't use the term the way your reference defines it.

Here is the standardized definition for "quantization"

quantization:
A process in which the continuous range of values
of an analog signal is sampled and divided into
nonoverlapping (but not necessarily equal)
subranges, and a discrete, unique value is
assigned to each subrange.

And you looked at a invalid definition for this context and
stated:

Note that in the definition, there appears no
mention of assigning a value. Assigning a value
would then be considered a part of a separate and
distinct process of converting to digital form, as
in

Obviously the Term of Art used in technical discussion
does indeed mean there *must* be a value assigned. In
fact it makes no sense at all unless that step is
included.

It seems there are a great many things that don't make sense to you.
If a "term of art" were to be defined in such a way that it
contradicts the definition for the exact same term as it is published
in Websters dictionary, I would be inclined to disregard it.

Then you go on to produce other, equally invalid in this
context, definitions for other terms of art:

"digital - of, or relating to data in the form of
numerical digits",

and as opposed to

"analog - of, relating to, or being a mechanism
in which data is represented by
continuously variable physical
quantities."

The first definition states that the form must be "of
numerical digits", and that is simply unnecessary. It
must be assigned a "value". The value can be numerical
digits, but it can be otherwise too. The point is that
it must be from a finite set of discrete values. (I
used the example of flags, where the value might be a
square flag, a round flag, or a triangular flag. No
numerical digits at all.)

As you can see, knowing how to use a dictionary is
vitally important...

Perhaps almost as important as realizing at which side of the analog
to digital convertor you're looking.

jk

 

[Floyd L. Davidson]

Real world signals have finite, not zero bandwidth. That means that they
have sidebands. As soon as any of the sidebands are at or exceed the nyquist

You meant 1/2 the Nyquist Rate. But the point is that
the statement above excludes all such signals. It
specifically says that the actual sampling rate will be
*above* the Nyquist Rate. There will not be any
aliasing at samplying rate which is higher than the
Nyquist Rate.

Therefore there *cannot* *be* any problem with aliasing.
There will, however, be quantization distortion.

rate, there is distortion of the wave because of aliasing. Therefore the
sample rate has to be somewhat higher than the sample rate of the signal,
which is characterized by its carrier frequency.

What???? The rate of the signal is characterized by its carrier
frequency? 1) What if there is no carrier? 2) What if there is
a carrier and there is an upper sideband?

Clearly the carrier is not what characterizes the signal
frequency or the sample rate.

Congrats to Freeman for avoiding making a different statement that avoids
the error made in the previous statement.

He said "at a rate equal to or higher than twice the
highest significant signal frequency". That is
precisely the same as the one you claim is wrong:

Nyquist's theorem:
A theorem, developed by H. Nyquist, which states
that an analog signal waveform may be uniquely
reconstructed, without error, from samples taken
at equal time intervals. The sampling rate must be
equal to, or greater than, twice the highest
frequency component in the analog signal. Synonym
sampling theorem.

Did you forget what you were arguing previously?

 

[Dan Coby]

Will you two just get a room. Please.

 

[Floyd L. Davidson]

It perfectly defines the term as I use it.

Yes. And in a technical discussion, you are off in Alice's
Wonderland.

I don't use the term the way your reference defines it.

But nobody else, unless they are also in Wonderland,
knows what you mean if you don't use standard definitions.
And in a *technical* discussion, that means the technical
term of art, not the common English term.

....

It seems there are a great many things that don't make
sense to you.

Why people want to make absurd claims about terminology
is one of them. Maybe you can explain that...

If a "term of art" were to be defined in
such a way that it contradicts the definition for the
exact same term as it is published in Websters
dictionary, I would be inclined to disregard it.

It isn't a contradiction. The term of art is usually a
stricter definition. It is certainly going to be more
rigorous.

Regardless, to ignore it is abject foolishness. Can you
imagine a lawyer in court ignoring all of the term of
art definitions in favor of the standard common English
dictionary definitions? The other side would be
ecstatic... and the judge would probably throw the
lawyer out the door!

If you don't want to use the technical terminology, why
not just step out the door and avoid all technical
discussions? You won't make sense, it it does become
annoying.

Perhaps almost as important as realizing at which side
of the analog to digital convertor you're looking.

Exactly. Now, if you don't want to use terms in a what
that will be understood, how can you even tell which
side is which?

You'll end up making grossly trivial errors the way
Jerry Arvins and a couple of others here do...

 

[Floyd L. Davidson]

Will you two just get a room. Please.

I believe that we already have one. Right here.

 

[Jerry Avins]

All qualified practitioners will recognize that you are wrong, and
obviously unqualified.

It is in fact a correct statement. Do you know what quantization
distortion is?

Yes. How is it related to sampling rate? Why would sampling exactly at
the Nyquist rate be adequate if the quantization were fine enough? Don't
you read the crap you cut and paste?

Don't answer. I'm through.

Jerry

 

[Mr.T]

Although, unless the original signal is a squarewave, saying "the
sample contains all the information of the original signal" may be
overly optimistic.

Did you not understand the words "band-limited signal",
or the words "sampled at a rate equal to or higher than
twice the *highest significant signal frequency* " ?

And I guess you have no idea how Fourier analysis applies to square waves
either?

MrT.

 

[[email protected]]

Audio IS mechanical so something has to move.

Then how is it possible for a digital audio device [like an iPod] to
play audio back w/out moving a disc or tape? There is a digital chip
storing info and playback does not require moving anything. Just how
is this possible? If it works in digital, why can't it work in

analog?

What is 'moving' isn't the tape, it's the address counter in the
playback RAM. If you stop the address counter you will get exactly the
same result as if you stopped the tape or the turntable -- nothing.
The address counter is most certainly in motion - if you change the
rate it counts, the pitch of the output changes. It just isn't
directly visible without a logic analyzer. If you make it count
improperly, it can mimic the stylus skipping on a record. Just as you
can run out of tape, you can run out of addresses.

The way a tape recorder works is to constantly replace the magnetic
'core' of the transformer (the head) by dragging it away as tape.

You know too many buzzwords but not what they mean or imply. Go read -
a LOT.

GG