Jerry, this is an interesting topic indeed. It first caught my attention 
some years back when someone made an argument in which they implied that 
a particular signal was not just digitized, but was PCM, as if it were 
some kind of special encoding; and I had to ask myself, is it not just 
plain digital data (and what they heck they mean my PCM).

Anyway, I don't know the exact origins, and it's easy to come up with a 
half dozens plausible explanations that *could* be right, but I'll just 
make a couple of observations:

Modulate means "to change"--so it's pretty easy to make a lot of cases 
for change in some manner. If you have digital data in some form (
decimal floating point values in a spreadsheet), and move it to another 
form (fixed point values in binary in computer memory), you have a 
modulation of some sort.

Also, the sampling process itself is a modulation. One model of sampling 
is to take an analog signal (we usually bandlimit it first), and 
modulate it by multiplying it by a unit pulse train. This gives us a 
modulated pulse train. We then store the height of each pulse by 
encoding it as a numerical value (the "code" part?). The result is our 
PCM representation of the analog signal. You could say that the data is 
our string of pulse codes, from the pulse modulation; the pulse codes 
can be converted back to pulses and run through a low pass anti-imaging 
filter for analog reconstruction later.

Again, this is something I made up--I don't know if this is the true 
origin of the term, but it seems reasonable.


In <ObGdnZw7CNMbQ9_eRVn-ig@rcn.net> Jerry Avins  wrote:
> Steve Pope wrote:
>> I'm not sure I understand the question.  Any analog signal
>> passed into an A/D creates a PCM version of that analog
>> signal, and the A/D can be more than one bit wide.
> 
> That's just a pulse code. There's no modulation.
> 
> Jerry

"Jerry Avins" <jya@ieee.org> wrote in message 
news:sMydnckE6OIQFN_eRVn-jg@rcn.net...
> Can you cite any credible sources? (Credible is a step down from 
> authoritative.)

http://www.google.ca/search?hl=en&q=pcm+format

> There was a time when many took "communist" to mean any individual who 
> disagreed with the speaker's politics, but that didn't make it so.

A tempting diversion, but still dodging the issue -- Y'all should be more 
civil to Radium.

--
Matt

"Jon Harris" <jon99_harris7@hotmail.com> wrote in message 
news:3aJ0f.6922$BU1.2844@trnddc06...
> "Steve Pope" <spope33@speedymail.org> wrote in message 
> news:dhv5e2$igc$1@blue.rahul.net...
>> Jerry Avins  <jya@ieee.org> wrote:
>>
>>> Steve Pope wrote:
>>
>>>> I'm not sure I understand the question.  Any analog signal
>>>> passed into an A/D creates a PCM version of that analog
>>>> signal, and the A/D can be more than one bit wide.
>>
>>>That's just a pulse code. There's no modulation.
>>
>> I can understand why that is what you may think, but
>> my statement above describes what is considered "pulse
>> code modulation".
>>
>> It's just a definition, it's arbitrary, but it's used
>> consistently in this way.
>
> Yeah, I always thought the "modulation" part was thrown on gratuitously in 
> the textbooks to make it seem like another in the family of PWM, etc.. 
> But to me, PCM always seemed like a different animal entirely.

It has essentially become a fancy marketing synonym for "digital". 

rhnlogic@yahoo.com wrote:
> Jerry Avins wrote:
> 
>>PCM stands for pulse code modulation. Can there be PCM without
>>modulation? Is it still PCM if the modulation carries more than one bit
>>per symbol?
> 
> 
> What a term or abbreviation, such as PCM, might mean is often domain
> dependent.  Various computer, consumer electronics, and multimedia
> "standards" groups have often (re)defined a term when used inside their
> particular portion of an industry differently from the usage in other
> fields.  So what I say below is probably completely irrelevant to
> consumer audio labeling.
> 
> I go back to the days when AM, FM, PWM, PAM, PPM and PCM all referred
> to various methods of transferring information.  PCM signals were
> differentiated from P?M in that the data was somehow encoded (via
> some table or time sequence usually, but not always, consisting of
> "ones" and "zeros"), and thus not directly related to the duration,
> amplitude, position or other analog demodulated characteristic of
> the signal.
> 
> Unlike another poster, I don't think an A/D was necessary for a
> signal to be PCM.  The data to be transfered could have started out
> completely in the digital domain (from a teletype keyboard for
> instance), thus requiring no conversion from analog.
> 
> Since the information to be transfered was "coded" it could pretty
> much represent anything (encrypted streams, ASCII or EBCDIC, for
> instance.  Maybe even morse code?)
> 
> And once the information is stored, it doesn't seem to have anything
> to do with "pulses" or "modulation", just coding.

Logic seems to support this minority view. :-)

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

"Steve Pope" <spope33@speedymail.org> wrote in message 
news:dhv5e2$igc$1@blue.rahul.net...
> Jerry Avins  <jya@ieee.org> wrote:
>
>> Steve Pope wrote:
>
>>> I'm not sure I understand the question.  Any analog signal
>>> passed into an A/D creates a PCM version of that analog
>>> signal, and the A/D can be more than one bit wide.
>
>>That's just a pulse code. There's no modulation.
>
> I can understand why that is what you may think, but
> my statement above describes what is considered "pulse
> code modulation".
>
> It's just a definition, it's arbitrary, but it's used
> consistently in this way.

Yeah, I always thought the "modulation" part was thrown on gratuitously in the 
textbooks to make it seem like another in the family of PWM, etc..  But to me, 
PCM always seemed like a different animal entirely. 

On Tue, 04 Oct 2005 17:41:34 -0700, Paulo Castello da Costa wrote:

> Jon Harris wrote:
>> "Jerry Avins" <jya@ieee.org> wrote in message
>> news:sMydnckE6OIQFN_eRVn-jg@rcn.net...
>> 
>>>Matt Timmermans wrote:
>>>
>>>> These days, PCM is the name given to the common uncompressed
>>>> representations of audio signals as bit sequences. Yes, I know that's
>>>> silly, but that's the way it is.
>>>
>>> If there's general agreement on the issue (there hardly seems to be),
>>> that's fine with me, but nobody told me about it. Can you cite any
>>> credible sources? (Credible is a step down from authoritative.) There
>>> was a time when many took "communist" to mean any individual who
>>> disagreed with the speaker's politics, but that didn't make it so.
>> 
>> I'm in agreement with Matt regarding the common use of PCM, at least in
>> the consumer audio industry. The term "PCM" in exactly the sense Matt
>> describes is used many times my home DVD player's documentation (JVC
>> XV-SA600). Some examples:
> 
> I don't think the examples are using it in the "common use" sense you
> mean. Let's look at them one by one:
> 
>> "There are some audio formats recorded on discs as shown below. Linear
>> PCM Uncompressed digital audio, the same format used on CDs and most
>> studio masters."
> 
> Here it's qualified as "Linear ... Uncompressed", to give it the "common
> use" sense you expect.
> 
>> "DVD with 192/176.4/96/88.2 kHz, 16/22/24 bit linear PCM"
> 
> Here again "16/22/24 bit linear ...".
> 
>> In fact, the digital output is labeled "PCM/Stream" (stream is used to
>> refer to Dolby Digital and other compressed/encoded formats).
> 
> I.e., the general term "Pulse Code Modulation", or "Stream" (of bits) is
> used to cover any encoded sequence of bits, not just linear.
> 
> PCM originally referred to the transmission of data samples as a stream
> of pulses chosen from a code set, as opposed to PWM (Pulse Width
> Modulation), where the width of each pulse would be proportional to the
> amplitude of the data sample, or PAM (Pulse Amplitude Modulation), where
> each pulse would be proportional to the amplitude of the data sample.
> The code set wouldn't need to be "zeros and ones", but could be
> something more complex, like "one, zero", "one, one", and so on (where
> each "zero" or "one" actually represents a pulse shape). It's all pretty
> logical.

I agree.  To elaborate:

The important point about "PCM" (any of the forms) is that it is fully
wide-band.  There is one independent code word per sample period. (As is
the case with PWM and PAM, of course.)  ADPCM is also one code per sample,
but the code is dependent on the preceding samples, and so typically has
frequency-dependent limitations. DD/AC3, DTS, MP3 etc are much more
complicated encodings, generally with much more complicated
frequency-dependent limitations, and generally not one code-word per
sample period.

Well, that's my understanding.

-- 
Andrew

Jerry Avins wrote:
> PCM stands for pulse code modulation. Can there be PCM without
> modulation? Is it still PCM if the modulation carries more than one bit
> per symbol?

What a term or abbreviation, such as PCM, might mean is often domain
dependent.  Various computer, consumer electronics, and multimedia
"standards" groups have often (re)defined a term when used inside their
particular portion of an industry differently from the usage in other
fields.  So what I say below is probably completely irrelevant to
consumer audio labeling.

I go back to the days when AM, FM, PWM, PAM, PPM and PCM all referred
to various methods of transferring information.  PCM signals were
differentiated from P?M in that the data was somehow encoded (via
some table or time sequence usually, but not always, consisting of
"ones" and "zeros"), and thus not directly related to the duration,
amplitude, position or other analog demodulated characteristic of
the signal.

Unlike another poster, I don't think an A/D was necessary for a
signal to be PCM.  The data to be transfered could have started out
completely in the digital domain (from a teletype keyboard for
instance), thus requiring no conversion from analog.

Since the information to be transfered was "coded" it could pretty
much represent anything (encrypted streams, ASCII or EBCDIC, for
instance.  Maybe even morse code?)

And once the information is stored, it doesn't seem to have anything
to do with "pulses" or "modulation", just coding.


IMHO. YMMV.
-- 
rhn A.T nicholson d.O.t C-o-M

Jon Harris wrote:
> "Jerry Avins" <jya@ieee.org> wrote in message 
> news:sMydnckE6OIQFN_eRVn-jg@rcn.net...
> 
>>Matt Timmermans wrote:
>>
>>> These days, PCM is the name given to the common uncompressed 
>>> representations of audio signals as bit sequences. Yes, I know
>>> that's silly, but that's the way it is.
>>
>> If there's general agreement on the issue (there hardly seems to 
>> be), that's fine with me, but nobody told me about it. Can you cite
>> any credible sources? (Credible is a step down from 
>> authoritative.) There was a time when many took "communist" to mean
>> any individual who disagreed with the speaker's politics, but that
>> didn't make it so.
> 
> I'm in agreement with Matt regarding the common use of PCM, at least
> in the consumer audio industry. The term "PCM" in exactly the sense 
> Matt describes is used many times my home DVD player's documentation 
> (JVC XV-SA600). Some examples:

I don't think the examples are using it in the "common use" sense you 
mean. Let's look at them one by one:

> "There are some audio formats recorded on discs as shown below.
> Linear PCM Uncompressed digital audio, the same format used on CDs
> and most studio masters."

Here it's qualified as "Linear ... Uncompressed", to give it the "common 
use" sense you expect.

> "DVD with 192/176.4/96/88.2 kHz, 16/22/24 bit linear PCM"

Here again "16/22/24 bit linear ...".

> In fact, the digital output is labeled "PCM/Stream" (stream is used
> to  refer to Dolby Digital and other compressed/encoded formats).

I.e., the general term "Pulse Code Modulation", or "Stream" (of bits) is 
used to cover any encoded sequence of bits, not just linear.

PCM originally referred to the transmission of data samples as a stream 
of pulses chosen from a code set, as opposed to PWM (Pulse Width 
Modulation), where the width of each pulse would be proportional to the 
amplitude of the data sample, or PAM (Pulse Amplitude Modulation), where 
each pulse would be proportional to the amplitude of the data sample. 
The code set wouldn't need to be "zeros and ones", but could be 
something more complex, like "one, zero", "one, one", and so on (where 
each "zero" or "one" actually represents a pulse shape). It's all pretty 
logical.

Paulo

Jerry Avins  <jya@ieee.org> wrote:

> Steve Pope wrote:

>> I'm not sure I understand the question.  Any analog signal
>> passed into an A/D creates a PCM version of that analog
>> signal, and the A/D can be more than one bit wide.

>That's just a pulse code. There's no modulation.

I can understand why that is what you may think, but
my statement above describes what is considered "pulse
code modulation".

It's just a definition, it's arbitrary, but it's used
consistently in this way.

Steve

"Jerry Avins" wrote ...
> Are uncompressed data transmitted with more than one bit per symbol -- 
> high-speed modem traffic, for instance -- PCM or not?

Doesn't quadrature modulation allow several bits per
baud, something around 4-5?