IS EVERYTHING DEBATABLE, REALLY?

[SoundAndMotion]

14 hours ago, kumakuma said:

 

 

Fixed that for you:

 

 

[SoundAndMotion]

10 minutes ago, mansr said:

That's not the waveform you'd get. Not at all. Whenever the magnetisation changes, the read head generates a positive or negative pulse depending on the direction of the transition. Contrary to what that drawing suggests, real hard drives probably use some variant of NRZ coding.

Thanks for the correction! But the voltage is analog and includes noise, right? It is NOT: 10100101

[SoundAndMotion]

2 minutes ago, mansr said:

So what? Once the signal has passed a pulse detector, demodulation, and error correction, the precise waveform coming from the read head no longer matters.

 

So! After the pulse detector, demodulator, and error correction, the signal coming out of the circuit looks more like my fake trace... put it on a scope and you’ll see analog voltages, not a series of 0’s and 1’s

[SoundAndMotion]

On 11/8/2017 at 1:06 AM, Ralf11 said:

 

 

Seems not everybody is getting the message, so here it is again.  [my emphasis]

 

Ah, but the physical representation does matter… just must less so than with analog.

 

The beauty of digital communication is that since the data can have only one value or state at a time from a small, finite number of values, it can made far less sensitive to noise and linearity requirements can be dramatically loosened.

 

With analog there is a (theoretically) infinite number of values, and we have to recognize that 1.001V is different from 1.000V. This is no problem as long as all previous circuitry is sufficiently linear (easy!) and the noise is at least 60 dB down (again easy). But distinguishing 1.000001V from 1.000000V requires care in the design of the circuits.

 

With digital, we replace exact (or nearly exact ) values needed for analog with ranges of acceptable values:

 

 

So with TTL, 2.8V is the same as 4.2V, i.e. it represents “High” or 1 or True. Also, we should notice the any analog voltage between 2V and (nominally) 5V, is fine, according to the table above. Actual datasheets for specific circuits must be checked for certainty.

 

This makes transfer, storage and distribution of digital data (e.g. music) must less prone to error, compared to analog. Since we add error detection and correction to these methods, errors are truly rare. But this assumes all designs stay within spec, and therefore the physical representation DOES matter!

 

So what’s the big deal? If we can guarantee that no noise from the digital side, of any kind, infiltrates the analog circuity anywhere from the DAC and beyond, there is no big deal. It does not even need to be “no noise”, simply an inaudible amount.

 

But we know that such perfect noise isolation does not always exist. Many try to throw in disclaimers that the DAC (or subsequent analog circuits) must be a “reasonable” DAC, a “modern” DAC, a “well-designed” DAC or any “half-way decent” DAC. But just because the methods for creating such DACs are “known”, it does not mean all that are produced use these methods. There are easy-to-find counterexamples.

 

I’m sure most of the “bits are bits” crowd know all of this already. I find their argument that the “physical representation doesn’t matter” to be hyperbole.
 

 

[SoundAndMotion]

I also wanted to add that the terms "digital electronics" and "analog electronics" relate more to the intended use than the way the components and sub-circuits work. Sure the transistors are optimized for the needed function in terms of speed, gain and size.

But I can build logic gates from analog op-amps and I can build a low-level amplifier from "digital" gates, the easiest being a digital CMOS inverter.

[SoundAndMotion]

3 minutes ago, mansr said:

You have no idea what you're talking about.

 

Care to back that up? I thought you were an engineer...? Prove it with an answer. Where do you disagree?

[SoundAndMotion]

2 minutes ago, mansr said:

I suggest you read a primer on electronics instead.

 

Well that was predictable! You are being lazy, because I know you are not stupid.

 

Okay, I already have read such books... several, and built many relatively simple analog and digital circuits still in use.

 

What are you having a problem with? Op-amps as gates or CMOS inverters as amplifiers? Should I find and paste some schematics for you? ... or you can google it yourself. Give me some time and I'll find a nice primer for you. I think you can find this stuff in Horowitz and Hill, but I'd have to check...

[SoundAndMotion]

41 minutes ago, mansr said:

I'm not lazy, but being your private tutor isn't my job.

 

Horowitz and Hill is a good introduction. Are you saying you've read it and still don't understand the difference between analogue and digital circuits?

 

I don’t need a tutor, but if I did, no worries, I won’t ask you. I’m doing just fine with my work.

But I do ask that your posts have content, not just empty, lazy swats at me. Either point out that I’m mistaken (and where!) or state that you disagree (and why!)

 

If you have no content to add other than just “nope” (admittedly, you usually do add useful content), either don’t answer me or I, too, can lazily just answer you “yup”.

 

I understand “digital” and “analog” pretty well. The different components used to construct a circuit with a specific intended use are pretty much the same. Or maybe you can tell us which transistors are digital-only and which are analog-only. Sure chips have been created and marketed for specific uses, but if reasonable sub-circuits are available, a smart and creative person (like you, I think, maybe I’m wrong) can “break the rules” and use them otherwise, as needed. But once you get to more complex chips or complete, specific circuits, torturing them to do something else becomes more difficult. But the labels “digital” and “analog” are the least of your worries.

 

Using an FPGA alone as a voltage regulator is tough, but not because it is “digital”.
Using an old-style analog computer as a binary to BCD decoder is tricky, but not because of the word “analog”.

 

EDIT: Yes, I have and have read H&H and agree with what FAS42 wrote above.

[SoundAndMotion]

9 minutes ago, mansr said:

That's where we disagree.

 

So your opinion is that I don't understand digital and analog differences pretty well, but you really don't say why. Should I take a test? (No, I'm not asking you to ask a lengthy essay question saying "Explain XYZ") Your opinion is noted, but I find you to be wrong. So does my success in designing, building and using digital and analog circuits.

 

9 minutes ago, mansr said:

Where did I ever suggest anything of the sort?

 

You didn't, but you also didn't find anything I said that was wrong... other than having an opinion different from mine. I was fishing for an understanding of your criticism, and explaining where some misunderstandings could occur...

 

18 minutes ago, mansr said:

What about this is so difficult for you people to grasp?

 

I understood and already knew all you wrote before you wrote it, i.e. I agree and learned nothing from the repetition. I think this explains a lot about your replies to me. Who is "you people"? And what  do you think "we" don't grasp? All you've explained does not contradict anything I've written. What is not grasped? Except perhaps semantic choice of word usage, that has little to do with understanding. We disagree on when/how to apply the labels "digital" and "analog". Is that the whole thing? Surely you don't support the idea that one find 0's and 1's running through digital electronics.

[SoundAndMotion]

1 hour ago, mansr said:

Then why do you keep pushing this ridiculous notion that digital is really analogue?

 

Apologies if I have failed, but I try to use “analog” and “digital” as modifiers, not as free-standing concepts. And in the context of this forum, we ALWAYS want to end up in an analog domain to hear the music. Therefore the presence of noise voltages (I hope we can agree this is an analog signal) anywhere in the system (yes, even on top of digital signals), that subsequently find their way to the final analog signal is a problem, if audible.

 

Although noise so large that it takes the analog voltage out of the range defined to represent particular digital encoding is typically either detected and corrected, or it results in gaps. And happily this is rare. 
I don’t have personal experience with noise so large that it invades the final analog pathways of my system through some form of coupling (ES, MS or EM or through the common power supply connections), but this is certainly not only possible but has been measured in other systems. Maybe I’m deaf.

 

So when people say:

Quote

 

Digital communications and computers work because the physical representation of the digital data doesn't matter.

 

 

I have a problem with the inaccuracy of “doesn’t matter”.

That is all I meant to say.
 

1 hour ago, mansr said:

What does that even mean?

 

It means that some people condescendingly say “doesn’t matter” cuz it’s all only 0’s and 1’s, implying analog noise is irrelevant. And I’m saying you can find voltages representing 0’s and 1’s, but I agree with you that talking about 0’s and 1’s has no meaning, without understanding that analog voltages or magnetic fields or capacitive charge or ... must be in the correct range so the value or transition happens at the right time.