Very interesting article. I haven't had time to read it completely in depth, but it introduces a different way of thinking about the LP: that there is effectively quantization due to the molecular structure of the LP.

http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page2.html

[Apologies if this has been posted before - my site search didn't turn it up]

If this has credibilty, might be a good addendum to the Wiki.

It's somewhat meaningless IMHO. Just because a medium is composed of finite-size particles doesn't make it digital. Most importantly, the positions of the particles is continuous, as is the motion of the stylus. The discretization of the time and space domains is a requirement of a digital system, and he goes nowhere close to proving that.

Finally, the vinyl surface deforms elastically, in a continuous fashion, during playback. That torpedoes the DR calculation. Certainly, there is an upper bound on the dynamic range placed by the molecular characteristics of the vinyl, but I would expect it to be much higher than 110db. And it has nothing to do with digital.

I read an article, either linked to form here or Ars Technica, of someone talking about how LPs actually have a less effective "sample rate" than CDs because of the chipping that occurs in the PVC when the LPs are made. I know micrographs of LPs are (relatively) readily available, it shouldn't be too hard to figure out exactly how small the discrete marks are.

A digital system uses discrete (that is, discontinuous) values to represent information for input, processing, transmission, storage, etc.

Vinyl is most certainly a form of digital, even the tape used to record the music has discrete values, and the recording head, the magnetic fields in the vinyl pickup and vinyl cutter are discrete. Everything in the chain of audio has discrete values, even those pesky little electrons turn out to be discrete.

And I’m sure you know that those ears of your are anything but true analog, there is discrete information going from your ears to the brain (and other senses too) all the time.

Paul



An analog world is an illusion of the brain.

And whoops, the terms become completely meaningless.
I tip my analog hat at you, dear sir.

Yeah Paulhoff, are you trolling, or do you seriously believe that BS?

What BS, are you trolling and believe in your BS.

Paul

Assuming that this is still a serious topic...

"Analog" or "digital" is really a classification determined by the type of media AND the method used to interact with it.

If you put a stylus in an LP groove, you're going to get an analog signal out of it. If you photograph it with an electron microscope, you'll get digital values back. They won't be suitable for passing to a preamp...

If you pass a magnetic tape over a gap in a solenoid, you'll get an analog signal. If you paint the tape's surface with certain types of dye, you'll see the magnetic domains in all their digital glory.

Examine a CD at an angle with bright light and you'll get a very nice continuous flow of rainbow colors. Look at it with a 700nm laser and you'll see a bunch of discrete elliptical holes.

In general, if you look at anything closely enough it'll appear discrete in structure. Whether it makes any sense to do so for the intended purpose is a separate question.

True analog is not on a vinyl record, it does not have infinite frequency band width, it does not have infinite dynamic range and it does not infinite time, therefore it can be quantified in a digital manner.

Paul

Yes, but if you look closely enough, nothing is quantifiable. At least, according to Heisenberg .

Therefore, everything is analog. QED.

-brendan

Paul does have a point, here. At the hairy edge of physics, everything is both discrete and probabilistic.

The whole idea of representing vinyl as sampled, however, is totally bogus (and Paul isn't saying anything otherwise, either), because random-sized bits of vinyl do not create periodic sampling. What's more, at high frequencies, the "wiggle" in the groove isn't even what holds the information, the high frequency information is created by density variations in the vinyl due to pressing, and the pressure of the stylus is what "reads" them by deforming the vinyl in an elastic (hopefully) fashion.

People have often calculated the "information content" in vinyl, which is entirely possible, because Shannon's (IT) Theorem applies entirely to any signal, discrete or pseudo-continuous, and it is in fact less than what you see in a CD. There is in fact more bandwidth, but the very low SNR more than compensates for the higher bandwidth.

One has, in the past, often seen a gob of bad science wherein it is stated that a vinyl record does 20Hz to 40kHz at 85dB. This is bogus for more reasons than one can imagine. First, the 85dB is C message weighted, and that means that all HF and LF noise is rejected. The real, applicable SNR is more in the 60dB range for the best vinyl under the best situations with the best playback system. Applying weighting is not valid, and pretending that one can put huge levels on vinyl at high frequencies is likewise not valid.

After that, the peak value as a function of frequency varies greatly on an LP. So you must further account for that, when you try to calculate SNR at a given frequency. After THAT, you can't count 2 independent channels much over 1kHz, so you have to stop multiplying by 2. After all is said and done, there is a respectable Shannon bound for information on an LP, but it's well under that 1.4114 mb/s on a CD.

I've also seen some outrightly ridiculous claims that "the sampling rate of vinyl is the size of an atom".

Vinyl enthusiasts will do anything to try to excuse the fact that they like the distortions on vinyl. In fact, implimenting such distortions (the polynomial ones, not the noise and pops and clicks) in digital makes digital sound better, too, to many people. So it goes. But it's not "more information". Feh!

Btw, I happen to know Paul is a CD enthusiast. When you read what he's saying, bear in mind that he has the physics to support it, but that he might, just might, be pointing out that the interpretation offered in the OP is a widdle bit astray...




Now, now, Heisenburg just says you can't tell where the discrete quanta ARE at the same time you know where they're going. They're still discrete quanta.

delta X * delta P >= h/2

Where Planck's constant is pretty small! So even the quantization is well below the information levels we're trying to extract from the medium. Not disagreeing, just trying to be precise.

Best known value of h is something like 10^-34 Joule-sec. whereas the classical atomic radius is about 10^-10 m; I'll bet the mass of the stylus is maybe .01 Newton and the velocity about 1/3 meter/sec.. That's a limit on uncertainty that won't hinder accurate reading of CD pits, let alone prevent us from treating that massive chisel sitting in the trench as an analog system.

Again, just thought actual numbers might give a sense of perspective to the 'uncertainty' situation.

Well, I think mass of the stylus in milligrams would be more appropriate than mass in Newtons, yes?

But I certainly don't disagree with your main point.

As to CD pits, the size is actually chosen to interfere with the laser wavelength, that's a different mechanism altogether, of course.

Absolutely - but then I'd have to throw in a scale factor anyway; I was trying not to deviate too far from the form of the equation for fear of making my post less comprehensible.

Yup, the depth of the pit is chosen so that the laser reflection from the bottom is nearly cancelled by interference. I was thinking more along the lines of perceiving/seeing the pits themselves, which are longer/wider than they are deep. But I'm already stretching my analogy pretty far.

No. It is not. It depends ONLY on the method.

You know, to me, that is a very scary thing, that shows another layer in the chain of distortions in vinyl records that some love so well for all the wrong reasons.

Paul

I don't know if it's uselss or not, but since I bought my first CD Player back in the late 1980s, I NEVER had any interest in the sound of vinyl again - pops, clicks, distortion, speed issues, more distortion at the last track of a side, jumps, noise, dust. Vinyl was doomed to death, and I don't think it completely died because a couple of freaks who want to make a big deal out of it...

I have no problem with people who like vinyl because of the distortions, as long as they know that there are problems and if they still like the sound the vinyl in spite of them, that is being objective and honest. I do have a problem with so-called objective claims about vinyl that aren't honest.

Paul

Can this just be settled by agreeing that vinyl is somewhat of a 'quantum' storage format? Both, quite arguably from what can be seen in this thread, digital and analogue all at once...

... I like to think of it that way, at least

Very interesting thread, made me think a little differently about one of my favourite audio storage formats.

No. But a "digital vinyl" is possible (but not easy to implement.) In this case it would be as good as any other digital format, without clicks and pops.

(Puts on Pedant Hat)

Um, what we usually call "analog" is in fact a continuous time, continuous-level analog of an electrical signal.

What we usually call "digital" is a sampled time, quantized level ***ANALOG*** of an electrical signal.

They are both analogs. Its the quantized time AND level that makes "digital" reproducible without copying error, when you get down to it.

While you are being pedantic, how would your "analog" versus "digital" scheme classify PWM, PPM, PAM, etc, which are sampled time, continuous level analogs?

The "digital analogue" versus "analog analogue"* distinction is arbitrary anyways. As you point out, discrete-time vs. continuous time and quantized vs. continuous level are more sensible categories. Using this classification we have:
discrete, quantized: PCM, DSD, etc.
discrete, continuous: PWM, PPM, PAM, etc.
continuous, continuous: Vinyl, tape, etc.
continuous, discrete: Can anybody think of an example here?

Of course, the only difference between quantized and continuous level is the shape of the noise floor, which is what this "vinyl is digital" debate seems to be about. In vinyl, the "random" noise floor is far far above the atomic level quantization noise floor, so there is no doubt in my mind that vinyl fits logically into the continuous-time continuous-level category. Most sensible people would call that "analog", in the audio sense.

* Yes, that is a serious abuse of the English language

The discrete-level, time-continuous has been handwaved at, and you might argue that PWM contains some vestige of it, but in general, it's the least common. It's been tried in modems, but appears to offer no advantage not also contained by PCM.

Consider the most simplified (and mostly useless) form of PWM, wherein the "PWM" is simply a comparitor that tests for >=0.

I grant you it's useless, and even a multilevel version isn't known to offer anything not offered by PCM in the real world, at least. We can discuss the characteristics, but the question 'why' enters rapidly.

How about using a confocal microscope instead?

http://irene.lbl.gov/