At what point does your circuit become "digital"?

[EEVblog]

Zen of Digital:
Digital exists as long as you think of it in terms of 1 and 0.
The closer you look at something you consider digital, the more analog it becomes.
Digital is just a frame of mind, the reality is that everything is analog.

The exact same thing can be said of analog.
Nothing is really "analog electronics", it's just a frame of mind, if you go deep enough then it's all really quantum physics.
Beyond that, it's turtles all the way down.

The art of engineering is often (or at a fundamental level) abstracting to a higher layer to get the job done.

[tggzzz]

The art of engineering is often (or at a fundamental level) abstracting to a higher layer to get the job done.

Very true.

I'm continually horrified at how many people only comprehend one or two layers of abstraction, and are fundamentally ignorant of other layers. That's even true when remaining entirely within software systems, let alone considering how they map onto hardware and/or could be implemented in hardware.

[Brumby]

.... like TCP/IP?

[tggzzz]

Oh, sick!

Good

So at a glance, this seems right:

Information theory is an important extra dimension, e.g. the concepts of entropy, error detection and correction, and state spaces don't have an analogue equivalent.

I was just contemplating these and considering counterpoints.  Offhand, entropy and state space are perfectly comfortable in the analog domain -- statistical mechanics seems an adequate, very low level example of the former, while state space control theory is perfectly comfortable with real-valued (or complex, I suppose) variables in the state vector (hey, it's just an array of differential equations, nothing digital needed).

Mumble mumble gripe groan ... drat, I have to agree. (In theory rather than in practice, though )

I suppose one could also, with a stretch of the imagination and terminology, consider analogue spread spread spectrum systems (e.g. FM ) as a form of error-correction without error-detection. No, I don't want to (attempt to) justify that.

But offhand, I didn't know of any examples of the emphasized case.

First hit on Google.  The headline and abstract seem to be on the money.  I will read this with great interest.
http://preserve.lehigh.edu/cgi/viewcontent.cgi?article=2035&context=etd

Thanks for inspiring this idea

That's a new one for me!

Skipping to P90, and then speed reading without too much comprehension, they note:

• "... Hamming distance, a key metric in digital codes .... to the adoption of related concepts such as maximum distance separability, Hamming distance is not nearly as indicative in the analog domain as in the digital domain.", P91
• "... We provide a geometric method to illustrate how and why, and point to nonlinear codes as the solution for analog error control.", P91
• "... As linear analog codes do not perform well on AWGN channels, we further study nonlinear analog codes.", P159
• "... We cleverly exploit the butterfly effect of chaotic systems and the successful concatenation idea from digital turbo codes, and propose two new classes of chaotic analog codes: chaotic analog turbo codes and mirrored baker’s map codes.", P160

I'm undecided as to whether that means they have fuzzy digital state spaces (boring), or they have found a practical benefit from using for chaotic systems (interesting), or both.

[tggzzz]

.... like TCP/IP?

That's merely one small instance of it.

By "it" I mean the context you omitted when you pressed "reply" rather than "quote"; this forum isn't edaboard nor stackexchange The context was  "I'm continually horrified at how many people only comprehend one or two layers of abstraction, and are fundamentally ignorant of other layers. That's even true when remaining entirely within software systems, let alone considering how they map onto hardware and/or could be implemented in hardware."

Start by considering an application using web services (especially SOAP and its allies). Then count how many abstraction layers there are between the application and voltages on networking cables. For entertainment, count how many times an asynchronous protocol is layered on a synchronous protocol or vice versa.

[Beamin]

Zen of Digital:
Digital exists as long as you think of it in terms of 1 and 0.
The closer you look at something you consider digital, the more analog it becomes.
Digital is just a frame of mind, the reality is that everything is analog.

The exact same thing can be said of analog.
Nothing is really "analog electronics", it's just a frame of mind, if you go deep enough then it's all really quantum physics.
Beyond that, it's turtles all the way down.

The art of engineering is often (or at a fundamental level) abstracting to a higher layer to get the job done.

I was just thinking about that right before I read this thread:
"What holds up the earth?"
- It sits on the back of a giant tortoise.
" And that sits on what?"
- Another tortoise: Its tortoise all the way down.

Isn't that from a roger penrose book or paul davies book?

Every thing is digital because you get down to plank length and plank time and teleporting of electrons. Did you know there are only 10^120 plank volumes in the observable universe and it is only filled with 10^80 particles? I think for the multiverse it would be ghrams number big. Or even just our unobservable size taking into alan guth theory where it 10^47 times bigger we might measure it in a 2 or 3 grams numbers in size.

Dave: Stop reading my thoughts before you post. Click the preview button and see if you are quantum entangling our thoughts together before you post at sub luminal speeds or at least collapse the probability function BEFORE posting.

[Harb]

All the lights in my house are Digital, bar one in the living room, which is analogue.

[buck converter]

When I build a circuit with logic gates, and explain it to people with less knowledge, in person, I call it analog, because had I said digital, people would think that it involved software.

[adras]

Well, I don't know much about electronics, but as a software developer I've got something in mind.

I'd consider digital circuits those which are solely based on binary logic. If something is one, another thing is zero. In analog circuits however there is not just 1 or 0, there's a whole range of values. Imagine charging up a capacitor, it's not either full or empty. It can be half full, or half empty which is important to consider. Or look at an oscilloscope, if you can measure a "clean" square wave I'd say it's digital, every other wave is analog.

[tggzzz]

Well, I don't know much about electronics, but as a software developer I've got something in mind.

I'd consider digital circuits those which are solely based on binary logic. If something is one, another thing is zero. In analog circuits however there is not just 1 or 0, there's a whole range of values. Imagine charging up a capacitor, it's not either full or empty. It can be half full, or half empty which is important to consider. Or look at an oscilloscope, if you can measure a "clean" square wave I'd say it's digital, every other wave is analog.

That is pretty good, but would be better if you said "...consider digital signals those...".

The problem is that digital logic gates are merely analogue circuits designed to spend most of the time saturated against Vcc or Gnd. It is usually possible to bias them so they operate in the linear region.

Consider also flip-flops, which most people think of as being in a 0 or 1 state. But when clocking and switching there will be a small window where the flip flop is deciding "which way to go". Hit that window and the flip-flop will enter a metastable state for an indefinite time. Some circuits rely on metastability, and the higher the proportion of time spent in the metastable state, the better the circuit is performing! (Software equivalent of metastability: a distributed two-phase ACID transaction failing).

Consider also logic simulators, in which each signal can, often, be assigned 9 different values.

[T3sl4co1l]

Yeah, the field is more general than binary: it's digital.  Binary is just the most convenient realization of digital electronics, while being fairly efficient: a consequence of 2 being close to e.  The only better whole number is base 3, but it requires more transistors to implement, than information capacity gained in the base and efficiency from the proximity to e.

Mechanical computers in base 10 (going back to Babbage's engine, among others) are still digital, and still obey the same conditions I laid out before.

Indeed, you're very very close to the idea of "digital".  You just need to go a few steps further, and consider what it means to be digital -- functionally.  Literally: if it quacks like a digital circuit, it is one.

Tim

[james_s]

There are a lot of good answers buried in here although I wonder if the guy is more confused now than when he first asked

The world is inherently analog, but I would describe a digital circuit as one in which we only concern ourselves with discrete states and transitions. Technically every digital IC is inherently an analog circuit but we can ignore that for the most part, we don't care what precise voltage is present on an input or output at any given time so long as it falls within a range defined as either high or low. Many circuits are fairly easy to classify as either analog or digital but there is plenty of overlap. In the end it doesn't really matter, it's just a classification.

[Beamin]

Well, I don't know much about electronics, but as a software developer I've got something in mind.

I'd consider digital circuits those which are solely based on binary logic. If something is one, another thing is zero. In analog circuits however there is not just 1 or 0, there's a whole range of values. Imagine charging up a capacitor, it's not either full or empty. It can be half full, or half empty which is important to consider. Or look at an oscilloscope, if you can measure a "clean" square wave I'd say it's digital, every other wave is analog.

So you're a capacitor is half empty kind of guy?

[Beamin]

Yeah, the field is more general than binary: it's digital.  Binary is just the most convenient realization of digital electronics, while being fairly efficient: a consequence of 2 being close to e.  The only better whole number is base 3, but it requires more transistors to implement, than information capacity gained in the base and efficiency from the proximity to e.

Mechanical computers in base 10 (going back to Babbage's engine, among others) are still digital, and still obey the same conditions I laid out before.

Indeed, you're very very close to the idea of "digital".  You just need to go a few steps further, and consider what it means to be digital -- functionally.  Literally: if it quacks like a digital circuit, it is one.

Tim

So why e? For the mathematically illiterate. 

[T3sl4co1l]

Yeah, the field is more general than binary: it's digital.  Binary is just the most convenient realization of digital electronics, while being fairly efficient: a consequence of 2 being close to e.  The only better whole number is base 3, but it requires more transistors to implement, than information capacity gained in the base and efficiency from the proximity to e.

Mechanical computers in base 10 (going back to Babbage's engine, among others) are still digital, and still obey the same conditions I laid out before.

Indeed, you're very very close to the idea of "digital".  You just need to go a few steps further, and consider what it means to be digital -- functionally.  Literally: if it quacks like a digital circuit, it is one.

Tim

So why e? For the mathematically illiterate.

https://math.stackexchange.com/questions/446664/what-is-the-most-efficient-numerical-base-system



Tim

[fourtytwo42]

Something I think you may have all missed (I apologize if not) is the role of the dreaded PRODUCT MARKETING!

In my experience these people can call black, white if they think it will better attract the consumer.

There was a time before humble washing machines contained micro-processors that many goods that contained not a bit, were labelled digital. A bit like the Blue LED fad etc etc.

In certain quarters of the audiophile fraternity ANALOGUE is sought after and the presence of any digital components suppressed. 

[Awesome14]

I'm surprised no one got this one. A digital circuit is any circuit that converts analog to digital or digital to analog

edit https://www.elprocus.com/difference-between-analog-circuit-and-digital-circuit/.

 An analog circuit has no conversion to digital. For instance, a digital cell system samples the audio voice component and converts it into a digital quantity.

The cell phone itself converts the digital quantity back into analog and sends it to the ear speaker, and vice versa for the microphone. The reason for doing this is error checking. The cell phone receives digital packets + a checksum. If the cell phone's calculated checksum is different than the checksum sent by the cell system tower, the cell phone requests that the packets be sent again, and again, until the checksum is correct, which means the packets are intact, the information is correct, and the resulting audio signal will be correct.

With analog you can't do algorithmic error checking. 

Edit: you can do simple algorithmic error correction on analog circuits. In this paper link originally posted above: http://preserve.lehigh.edu/cgi/viewcontent.cgi?article=2035&context=etd

I'm not sure I would call the author's idea of analog error-correction the same thing. It seems more like fidelity-loss improvement. It uses sampling and conversion, but it's not linear, so he calls it analog.

[T3sl4co1l]

In my experience these people can call black, white if they think it will better attract the consumer.

Is the Moon black or white?



It looks "white" against the night sky, but its albedo of 0.12 suggests it's, at best, dark gray.

Context matters to visual determinations.

Regarding marketing; they're just providing that context.  Whether the context is realistic or imaginary, matters not. . .

Tim

[T3sl4co1l]

I'm surprised no one got this one. A digital circuit is any circuit that converts analog to digital or digital to analog. An analog circuit has no conversion to digital.

[citation needed]

With analog you can't do algorithmic error checking.

[citation needed]

In fact, I recently pulled up a cool article about this:
http://preserve.lehigh.edu/cgi/viewcontent.cgi?article=2035&context=etd

Tim

[Mr. Scram]

Well, I don't know much about electronics, but as a software developer I've got something in mind.

I'd consider digital circuits those which are solely based on binary logic. If something is one, another thing is zero. In analog circuits however there is not just 1 or 0, there's a whole range of values. Imagine charging up a capacitor, it's not either full or empty. It can be half full, or half empty which is important to consider. Or look at an oscilloscope, if you can measure a "clean" square wave I'd say it's digital, every other wave is analog.

Ternary computers exist, and something like charlieplexing uses the floating state of pins to do work. I would consider neither properly analog in nature.

[Mr. Scram]

I'm surprised no one got this one. A digital circuit is any circuit that converts analog to digital or digital to analog. An analog circuit has no conversion to digital. For instance, a digital cell system samples the audio voice component and converts it into a digital quantity.

The cell phone itself converts the digital quantity back into analog and sends it to the ear speaker, and vice versa for the microphone. The reason for doing this is error checking. The cell phone receives digital packets + a checksum. If the cell phone's calculated checksum is different than the checksum sent by the cell system tower, the cell phone requests that the packets be sent again, and again, until the checksum is correct, which means the packets are intact, the information is correct, and the resulting audio signal will be correct.

With analog you can't do algorithmic error checking.

You're basically saying that a digital circuit deals with digital signals. While true, it doesn't seem to be an answer to the question at all, because the difference between analog and digital is as murky as it was before the answer.

[vk6zgo]

Well, I don't know much about electronics, but as a software developer I've got something in mind.

I'd consider digital circuits those which are solely based on binary logic. If something is one, another thing is zero. In analog circuits however there is not just 1 or 0, there's a whole range of values. Imagine charging up a capacitor, it's not either full or empty. It can be half full, or half empty which is important to consider. Or look at an oscilloscope, if you can measure a "clean" square wave I'd say it's digital, every other wave is analog.

A square wave is a rectangular wave with a 1:1 mark/ space ratio ( 50 % duty cycle.)
There are many other types of rectangular waves--------in what way are they not digital?

OK, I'm being pedantic, but this is one of my pet hates.
Everybody calls any rectangular wave a "square wave", but it  is not the case.
You are a software person & don't know any better, but others do, & still perpetuate this sloppy usage of terminology.
Square waves are a special case of rectangular waves with a different (simpler) spectral distribution compared to other such waves.

[KL27x]

OK, I'm being pedantic, but this is one of my pet hates.
Everybody calls any rectangular wave a "square wave", but it  is not the case.

This might be a language issue. In many English speaking countries "square" is not just a specific polygon. It is also used as an adjective which means perpendicular, or even as an adjective to mean having 90 degree angles.

This could be one of those pet peeves you can let go of.

If you want to get really pendantic, voltage and time have no universal scale in relation to each other. Volts and seconds are arbitrary human units. Any "rectangular" half wave can be described as a 3-sided square (ignoring the rise/fall times) by tweaking the scale.

[T3sl4co1l]

I think he's referring to the class of rectangular wave which is self-inverse.  In other words, duty cycle = 50%, and a spectrum of odd harmonics that go as 1/N.

I tend to use "square" in the "right" sense (square corners), hence "squaring up" a signal (using an amplifier and clipper, or comparator stage), and neglecting rise/fall times in that sense.  It's more accessible (less of a mouthful and more universally understood) than "trapezoidal wave", which would be more strictly correct (but really, the corners are rounded over, but really the..).

Tim

[vk6zgo]

I think he's referring to the class of rectangular wave which is self-inverse.  In other words, duty cycle = 50%, and a spectrum of odd harmonics that go as 1/N.

I tend to use "square" in the "right" sense (square corners), hence "squaring up" a signal (using an amplifier and clipper, or comparator stage), and neglecting rise/fall times in that sense.  It's more accessible (less of a mouthful and more universally understood) than "trapezoidal wave", which would be more strictly correct (but really, the corners are rounded over, but really the..)

I have never heard  " square" used in the sense of square corners in over 40 years.
The only sense I have ever heard is a rectangular waveform with a 1:1 mark/space ratio (50% duty cycle)
(& yes, I neglected to change from 1:1 to 50% when I decided to use duty cycle instead of m/s ratio).

The spectrum of such a waveform is different from that of other rectangular waveforms of different duty cycle.
I recall having to work out the reason for the difference when I did the "retreads" course to become a Tech Officer with Telecom Aust.
The guys that ran the course were very hard on stuff like that.