Digital, or analog?

[Floyd L. Davidson]

COMPONENTS ;? :?
2 resistors ;/
Basically we agree :>

Yes, exactly! The two resistors are only a slight addition, but
of course they dramatically change the way the circuit works!

 

[Eric Jacobsen]

I was not referencing just *any* definitions. I say *those*
definitions. They are not in dispute by any reputable source.

You stake *your* reputation against the entire industry when
you deny they are correct, and that is simply absurd!


So we are going to ignore indisputably good standardized
definitions that the industry (not to mention the Federal
government) uses, and you think that is going to in any way
assist in discussing this topic???

Ron and I have both worked in standards bodies helping to write
standards, so we have a little different perspective on this.

A standard is typically intended only to reduce confusion in the
context for which it is intended.

e.g., the form factor for electrical wall outlets and plugs are
"standardized" around the world, but they're different standards. You
can go to Europe (or any of a host of places) around the world and
argue until you're blue in the face that "the standard" electrical
plug looks like the US version, and you can produce the drawings and
definitions to prove it, but you'll still not be able to plug in into
an outlet in the UK or Germany or Japan or China or Russia or a host
of other places. You'll be right in the context of the US, but it
won't help you run your shaver over there.

You insist on using a single definition that you pulled out of a
"standard" somewhere. It's causing you trouble and you don't seem to
be able to recognize that outside of the context for which that
standard was developed it's no longer "the" definition, just "a"
definition. Anyone else's definition is as arguably good here as the
one's you've cited, hence the current discussion.



Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

 

[Bob Myers]

I was not referencing just *any* definitions. I say *those*
definitions. They are not in dispute by any reputable source.

Well, not as long as YOU get to pick which sources are
to be considered "reputable" and which aren't...

Bob M.

 

[Bob Myers]

By definition, it implys that it is continous. Linear, no, but
continuous is essential.

Let's try another simple example, shall we?

Last year, the organizers of a charity drive came to their
local electronics-tinkerer type, and told him that they
wanted to put a big display up that would show how
well they were doing toward meeting their year's goal.

"No problem," he says. "I can rig up a big meter movement
to this here variable resistor, we'll put it on a billboard,
and I'll label various points on the meter as '10%,' '20%,'
and so on. Each day, you send someone to the billboard,
and they can turn this knob to set the pointer to wherever
it needs to be that day!"

"Perfect!" they replied, and the new charity-drive billboard
went up. Worked like a charm, too. Each day, the
designated knob-twister would adjust the position of the
pointer, and everyone was happy with the result. Until
THIS year's drive, when they found that the big potentiometer
that the tinkerer had originally used burned out, and no
replacement was available.

"Still no problem," he says. "I'll just replace it with
a string of twenty or thirty discrete resistors, and you
can then move the pointer by choosing where in that
string you attach this alligator clip. It'll work just like
before!" And sure enough, it did.

The question at this point should be clear. The position
of the pointer on this big charity-drive billboard is
being controlled by the current coming up to the
meter movement in both cases - and also, quite clearly,
when the thing was first put up that this current can be
considered to be an analog representation of how
close this group is to their goal. After the modification
was made, though, the thing still works in exactly the
same manner - the information is being delivered to the
billboard via a current whose amplitude is "analogous to"
that information. It just is no longer capable of being
*continuously* adjusted.

Are you arguing that we now have a "digital" signal
going over that wire? And are you going to actually
think about your answer, or are you going to defer all
thought to some "standard"?


Bob M.

 

[Bob Myers]

That is demonstrably incorrect. All circuits that have a physical
instantiation are analog, as are the signals that they process.

Well, it all depends on which definition of "analog"
you're using I guess. From the following comment:

[snip fairly lucid description of analog encoding]

....I would have assumed that we were in agreement
there. In which case "analog" and "digital" BOTH refer
to how information is encoded onto a signal, but not
really the signal itself. Elsewhere I have given an example
of a signal which cannot be unambiguously classified as
either "analog" or "digital" in the common (and sloppy)
use of those terms merely by looking at it. Given that
confusion, I am simply proposing that if you really want
to see what's going on, you think of circuits and signals
as simply being "electrical" - and leave the words "analog"
and "digital" for use only in pointing to encodings.

Admittedly, most of us in the industry are VERY used to
using these terms another way (and I do so myself in many
cases), and it takes a real effort to change that perspective.
But it is very often valuable to do so. Far too often, I have
seen various claims of advantages or disadvantages in a given
system presented as being the result of something "being
analog" or "being digital," when in fact these concerns had
nothing to do with what was being considered.

Bob M.

 

[Ron N.]

OK let's not tooooo far to the extreme .

So a relay is "analogue" in sense that on closure its resistance may
change "gradually" from 10^10 ohms thru 10^8 ohms to 10^-3 ohms.

Right. But my point was that the model has to be chosen
depending on the question. We can still model a relay as
digital because the elevators don't usually show any measurable
difference in behavior due to these analog realities. And we
can model a relay as analog if we are looking at, say, the
effects of humidity & back-EMF on the life of contact plating.


IMHO. YMMV.

 

[Eric Jacobsen]

Ron and I have both worked in standards bodies helping to write
standards, so we have a little different perspective on this.

Oop, sorry, Ron, if that's not true. I was mistaken over who's post
was included here.


Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

 

[Jerry Avins]

CAREFUL I out weigh you at least 3 to 2, even if decade younger

If you try to sit on me, I can squeeze through places you can't follow.

Jerry

 

[Jerry Avins]

I've seen a 7400 series inverter used as an 'op amp'
Don't recall why it was done, suspect was just to prove it could be ;/

Lousy choice. The symmetric CD40xx, on the other hand, was well enough
suited to be used in at least onecommercial product.

jerry

 

[Jerry Avins]

... In which case "analog" and "digital" BOTH refer
to how information is encoded onto a signal, but not
really the signal itself. Elsewhere I have given an example
of a signal which cannot be unambiguously classified as
either "analog" or "digital" in the common (and sloppy)
use of those terms merely by looking at it. Given that
confusion, I am simply proposing that if you really want
to see what's going on, you think of circuits and signals
as simply being "electrical" - and leave the words "analog"
and "digital" for use only in pointing to encodings.

Admittedly, most of us in the industry are VERY used to
using these terms another way (and I do so myself in many
cases), and it takes a real effort to change that perspective.
But it is very often valuable to do so. Far too often, I have
seen various claims of advantages or disadvantages in a given
system presented as being the result of something "being
analog" or "being digital," when in fact these concerns had
nothing to do with what was being considered.

I pretty much agree. It's too bad that some people become upset and
angry when their established mindsets are challenged.

Jerry

 

[Jerry Avins]

The faster the digital circuit goes the more analog it becomes.

That's a good sig!

 

[Scott Seidman]

To instantiate or process a digital
state, it must be encoded into a voltage/current analog.

True, but only well enough to differentiate a low from a high, and you can
also build in error correction.

 

[Steve Underwood]

Lousy choice. The symmetric CD40xx, on the other hand, was well enough
suited to be used in at least onecommercial product.

If there was a choice you'd use a proper op-amp. You only use these
gates because they are leftovers in a multi-gate package. 74xx devices
were used quite often in semi-analogue roles.

I find it an interesting paradox that in the old days nobody sold single
gate devices. Now simple logic has been reduced to the role of glue, a
number of suppliers make tiny one gate devices.

Steve

 

[John E. Hadstate]

That is demonstrably incorrect. All circuits that have a
physical instantiation are analog, as are the signals
that they process.

Well, it all depends on which definition of "analog"
you're using I guess. From the following comment:

[snip fairly lucid description of analog encoding]

...I would have assumed that we were in agreement
there.

Yes, in terms of your description of what analog means, I
think you were right on target.

In which case "analog" and "digital" BOTH refer
to how information is encoded onto a signal, but not
really the signal itself.

Depends on what you mean by signal. (This conversation is
becoming downright Clintonesque.)

Elsewhere I have given an example
of a signal which cannot be unambiguously classified as
either "analog" or "digital" in the common (and sloppy)
use of those terms merely by looking at it.

I read it. I thought it was brilliant.

Given that
confusion, I am simply proposing that if you really want
to see what's going on, you think of circuits and signals
as simply being "electrical" - and leave the words
"analog"
and "digital" for use only in pointing to encodings.

I started to make that distinction. Then I thought about
many examples of analog and digital hydraulic and mechanical
circuits and decided to go with the more general concept.

Admittedly, most of us in the industry are VERY used to
using these terms another way (and I do so myself in many
cases), and it takes a real effort to change that
perspective.
But it is very often valuable to do so. Far too often, I
have
seen various claims of advantages or disadvantages in a
given
system presented as being the result of something "being
analog" or "being digital," when in fact these concerns
had
nothing to do with what was being considered.

I think we are in violent agreement ;-)

 

[Jerry Avins]

Sure. You may think of a one-shot as a digital device, but it takes a
fair amount of analog savvy to build one whose pulse width is
independent of duty cycle.

If there was a choice you'd use a proper op-amp. You only use these
gates because they are leftovers in a multi-gate package. 74xx devices
were used quite often in semi-analogue roles.

I recall one audio device that used an entire hex inverter chip as a
stereo headset amplifier.

I find it an interesting paradox that in the old days nobody sold single
gate devices. Now simple logic has been reduced to the role of glue, a
number of suppliers make tiny one gate devices.

I may have a few one-gate RTL packages (six lead, round).

Jerry

 

[Ray Andraka]

Sure. You may think of a one-shot as a digital device, but it takes a
fair amount of analog savvy to build one whose pulse width is
independent of duty cycle.

As a digital guy, I sure don't consider one-shots digital. By
one-shots, I am referring to those devices that use the physical
characteristics of components (typically capacitance and resistance) to
set a time delay.

The distinction I use in my book (which sadly will probably never be
finished because of the time involved writing it) is that an analog
circuit depends on the physical properties of the components of the
circuit as an integral part of the processing. A digital circuit
processes the signal through a logical or numerical manipulation of the
signal so that the circuit itself is only incidental to the processing.

Clearly, with that definition, one-shots such as the 74123 belong to the
class of analog circuits regardless of whether they were offered as a
component within a digital logic family. The only thing digital about
these parts is the logic family compatible I/O and the simple logic
controls built in to allow the one shot to be operated by and output to
the logic family.

Another circuit that used to be commonly used with digital logic that is
really an analog part is the delay line.

 

[Jerry Avins]

As a digital guy, I sure don't consider one-shots digital. By
one-shots, I am referring to those devices that use the physical
characteristics of components (typically capacitance and resistance) to
set a time delay.

The distinction I use in my book (which sadly will probably never be
finished because of the time involved writing it) is that an analog
circuit depends on the physical properties of the components of the
circuit as an integral part of the processing. A digital circuit
processes the signal through a logical or numerical manipulation of the
signal so that the circuit itself is only incidental to the processing.

Clearly, with that definition, one-shots such as the 74123 belong to the
class of analog circuits regardless of whether they were offered as a
component within a digital logic family. The only thing digital about
these parts is the logic family compatible I/O and the simple logic
controls built in to allow the one shot to be operated by and output to
the logic family.

Another circuit that used to be commonly used with digital logic that is
really an analog part is the delay line.

I'd say that a one-shot is digital in intent and analog in execution. I
believe that's true of any physical digital component, just more clearly
evident with circuits that intentionally use Rs and Cs for timing. All
circuits incorporate Rs and Cs, whether intentionally or not. Memory
buses are "digital" too, but modern clock speeds dictate that they be
designed like the broadband transmission lines they really are.

Very few dichotomies are entirely clear. For most, it is possible to
assign most items to one class or the other although the matter is
ambiguous for some near the division. The analog-digital dichotomy puts
the fuzzy dividing line over near one edge. We may use current,
pressure, voltage, fluid flow and other analogs to represent ones and
zeros, but they are all analogs nonetheless. Those ones and zeros exist
in pure form only in out minds. Even marks on paper are analogs of a
sort. We move the division to match a particular purpose. Assuming that
it is fixed hobbles our thoughts.

I began this thread to isolate this discussion from the rancor
developing in another one. I'm pleased to note that it seems to have
succeeded.

Jerry

 

[Bob Myers]

Depends on what you mean by signal. (This conversation is becoming
downright Clintonesque.)

Well, THAT would depend on just what your definition
of "is" is....

But I think at this point we ARE in violent agreement about
pretty much all the real points, and are in danger of starting
a task of rearranging the semantic deck chairs, so I
think I'll quit until something else comes along.

I read it. I thought it was brilliant.

Thanks!

Bob M.

 

[Ray Andraka]

I'd say that a one-shot is digital in intent and analog in execution. I
believe that's true of any physical digital component, just more clearly
evident with circuits that intentionally use Rs and Cs for timing. All
circuits incorporate Rs and Cs, whether intentionally or not. Memory
buses are "digital" too, but modern clock speeds dictate that they be
designed like the broadband transmission lines they really are.

Very few dichotomies are entirely clear. For most, it is possible to
assign most items to one class or the other although the matter is
ambiguous for some near the division. The analog-digital dichotomy puts
the fuzzy dividing line over near one edge. We may use current,
pressure, voltage, fluid flow and other analogs to represent ones and
zeros, but they are all analogs nonetheless. Those ones and zeros exist
in pure form only in out minds. Even marks on paper are analogs of a
sort. We move the division to match a particular purpose. Assuming that
it is fixed hobbles our thoughts.

I began this thread to isolate this discussion from the rancor
developing in another one. I'm pleased to note that it seems to have
succeeded.

Jerry

Ah, but Jerry, I think perhaps you missed the distinction. The one-shot
depends on the properties of the R's and C's to set up the timing
parameter. In essence, the one shot is comparing a decaying analog
voltage against a reference analog voltage and outputting a digital
signal to indicate whether or not the reference threshold has been
crossed. There is clearly an analog component to this circuit that
would not work the same if the values of the components were changed.

Memory, and other digital circuits that incorporate capacitors and
resistors do not depend on the values of those components to set the
behavior of the circuit. The digital realization could be moved to
another logic foundation (say hydraulics) and it would perform the same
function (albiet, maybe not as fast). The analog circuit however
depends on the characteristics of the components, so moving it to
another technology generally means either finding a component with
equivalent characteristics or modeling the behavior of the replaced
component with something else that mimics the physical characteristics
of the component. See the fundamental difference is the digital is
performing a numerical or logical manipulation of the signal where the
analog is modifying the signal by subjecting it to physical properties
of the materials, which generally is not quantized.

Digital, by design quantizes the analog nature of the underlying
circuit to represent numbers, but fundamentally you are doing numerical
operations rather than relying on the physical properties of a component
to do the processing (decay of a voltage, for example).

 

[Jerry Avins]

Ah, but Jerry, I think perhaps you missed the distinction. The one-shot
depends on the properties of the R's and C's to set up the timing
parameter. In essence, the one shot is comparing a decaying analog
voltage against a reference analog voltage and outputting a digital
signal to indicate whether or not the reference threshold has been
crossed. There is clearly an analog component to this circuit that
would not work the same if the values of the components were changed.

Memory, and other digital circuits that incorporate capacitors and
resistors do not depend on the values of those components to set the
behavior of the circuit. The digital realization could be moved to
another logic foundation (say hydraulics) and it would perform the same
function (albiet, maybe not as fast). The analog circuit however
depends on the characteristics of the components, so moving it to
another technology generally means either finding a component with
equivalent characteristics or modeling the behavior of the replaced
component with something else that mimics the physical characteristics
of the component. See the fundamental difference is the digital is
performing a numerical or logical manipulation of the signal where the
analog is modifying the signal by subjecting it to physical properties
of the materials, which generally is not quantized.

Digital, by design quantizes the analog nature of the underlying
circuit to represent numbers, but fundamentally you are doing numerical
operations rather than relying on the physical properties of a component
to do the processing (decay of a voltage, for example).

I wasn't thinking of the memory chips themselves, with charges on
capacitors standing in for ones and zeros, but rather the analog nature
of the interconnections, with their transmission delays, characteristic
impedances, and matched terminations. A flip-flop' set-up and hold times
arise from the underlying analog nature of its components. Even the
pulse width of the one-shot is quantized in trinary a way: long enough,
OK, and too short. Otherwise, they wouldn't be useful. (The memory
chip's charges are similarly quantized: in the zero range, trouble, and
in the one range.) I don't see the distinction between a digital and an
analog circuit element as fundamentally more meaningful than the
distinction between a digital and an analog wire.

Jerry