Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Vinyl is equivalent to which digital bit-depth and sampling rate? (Read 117552 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #100
But analog has infinite resolution!!!!1111111


That (as we both know) audiophile myth is a testimonial to widespread ignorance of Shannon's Information Theory even 66 years after the publication of his seminal paper:

Shannon, C.E. (1948), "A Mathematical Theory of Communication", Bell System Technical Journal, 27, pp. 379–423 & 623–656, July & October, 1948. PDF.

But it seems so intuitively clear... ;-)

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #101
If one is young (or even not so young, like me) and full of misconceptions, then one is very likely to ask questions that are based on false premises. Necessarily, one has to be prepared to take a knock or two in the process and, I believe, we should just be glad that there people willing to patiently explain and straighten us out. The threads are often useful, as the misconception of one must be the misconception of many.

Don't know how I would have felt about all that when I was "student-"aged: even 40-plus years on I can still pig-headed. I guess young Mr cdroid is just starting on this path where the learning is as much human as technical

Well stated. It's too bad he feels he was "bullied" and driven away from the forum. There's not much that can be done about that, though, if he isn't even willing to read and think about the responses and questions asked of him, and he's unwilling to consider the possibility that some of his knowledge wasn't entirely correct because either he misunderstood aspects of the information sources he holds in high regard, or the sources themselves were flawed.

It's not unusual to take great comfort in feeling very certain and very right about whatever we've learned, especially when the information came from sources we regard as authorities. But I think there's a real divide between people who, in situations where it's really no big deal, can let go of that certainty and its associated comfort, and those who can't. For topics like the nature of reality and what happens when you die, I can understand some people being very passionate and digging in their heels when their beliefs are challenged. But for something like analog/digital comparisons, and whether one can really hear a difference between this and that, it's shocking just how many people will allow that to be the hill they die on. If I learned tomorrow that everything I thought I knew about audio was wrong, I'd be like, "oh, wow, that's pretty crazy, but okay"—much like I did after doing some blind testing—whereas others, possibly even a younger version of myself, cover their ears and run.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #102
... Well stated. It's too bad he feels he was "bullied" and driven away from the forum. There's not much that can be done about that, though, if he isn't even willing to read and think about the responses and questions asked of him, and he's unwilling to consider the possibility that some of his knowledge wasn't entirely correct because either he misunderstood aspects of the information sources he holds in high regard, or the sources themselves were flawed. ...


Go over to the Purple Place and search for a thread started by cdroid in the last couple of days.
Regards,
   Don Hills
"People hear what they see." - Doris Day

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #103
Digital is not accurate except within its own domain.

Ding ding ding!  We have a winner.

Digital is infinitely accurate within its own domain.  That was my whole point in saying it's not comparable to analog.  What I did not say, however, is that it's infinitely precise.  Because it obviously isn't.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #104
As long as you stay digital you can pretty easily achieve arbitrary precision. If you used numbers as big as they are being used in cryptography, you'd have multiple thousands of decibels of SNR.

But the ~300 dB of 64-bit floating point ("double precision") is usually enough even for demanding DSP algorithms.
"I hear it when I see it."

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #105
I believe 64 bit float has a range of 300 decades, not dB.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #106
Significand precision is 64-11 = 53 bits.

20*log10(2^53) = 319 dB


300 decades would be 10^300, a 1 followed by 300 zeros.
"I hear it when I see it."

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #107
The smallest value that can be represented in 64 bit float is 2^-1024 or 10^-308.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #108
Precision and range are not the same thing.  DP has a range of about +/- 10^308 and a precision of 52 bits (about 16 decimal places or 320 dB).

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #109
In that case 64 bit integer (379 dB) would be a better use of the bits.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #110
In that case 64 bit integer (379 dB) would be a better use of the bits.
You only need that precision if you do lots of sums of large and small numbers, where the result would vanish, and that loss of precision might be noticeable later. So precision is not always the best choice if you just need (dynamic) range.
It's only audiophile if it's inconvenient.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #111
Digital is not accurate except within its own domain.

Ding ding ding!  We have a winner.


No, we have a recitation of generally accepted scientific truth.

Quote
Digital is infinitely accurate within its own domain.  That was my whole point in saying it's not comparable to analog.


Analog and digital are still entirely comparable. They can and frequently are used to do the identical same things.  The performance of either can be characterized using the identically same generalized measures.


Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #112
In that case 64 bit integer (379 dB) would be a better use of the bits.

Sure, floating point is a tradeoff. SNR for dynamic range. As long as you are within a certain range, 0 dBFS to roughly -66 dB, you are better off with integers.
Higher level and you get clipping. Lower level and the SNR will decrease.
"I hear it when I see it."

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #113
We use only integer in Rockbox (no floating point unit) and it is quite painful to maintain high snr.  You can of course do it but you must carefully consider the scaling of each operation to avoid excess rounding error. I do not recommend it if you can avoid it.

Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #114
Indeed, an integer DSP will often have something like a 40 bit MAC register when working with 32 bit integers for that reason.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #115
Jitter is comparatively easy to process out. Wow and flutter more difficult since there is no timing reference in the signal.

If the jitter happened in A/D, what is your timing reference?

Any known, steady signal from the analog input. Candidates include: Power line hum, switchmode power supplies, video displays, HVAC, etc.

Celemony's Capstan  (http://www.celemony.com/en/capstan ) is a $5K product that claims to be able to exploit such things.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #116
Coming back to the original topic of this thread, I have some questions.

I've seen the 65 dB figure around and I can definitely agree on that, having looked at the noise of some needledrops. However, how good or bad does it get and within what tolerance of a "max signal" are they referenced? For example, I've seen in the manuals of various tape recorders a lot more standardization with regards to these tolerances. The S/N ratio is calculated at a fixed reference, or explicitly a number of dB above the reference. (they may note their S/N is at reference, or at +3 dB above it, etc.) Additionally, I've seen some figures for third-harmonic distortion for different signal levels relative to reference.

Are there any as-detailed looks at specs on lathe cutting or are there too many more variables? (cutting, THEN the performance of the playing needle!) If there a reference amplitude for disc cutting? Any figures for distortion as it relates to level above any such reference? And then signal-to-noise ratio relative to that reference?

Separately, I'm curious about the 18 kHz figure for frequency response. With what reference is this chosen to be the max as well? Some amount of distortion or some amount of roll off? I've seen spectrograms which show distortion products creating 'false' high-frequency content, but does anyone know of any more precise tests? Any pure HF tones cut into vinyl? Maybe sweep up to 20 kHz and see how the original fares vs the added distortion?

If anyone has any materials I'd love to see them!

Oh, and does anyone know of the type of noise vinyl records exhibit? (or maybe the record plus the preamp?) Is it pink? Brownian? It seems like it's one of them, but turns white past some point.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #117
Quote
The S/N ratio is calculated at a fixed reference, or explicitly a number of dB above the reference. (they may note their S/N is at reference, or at +3 dB above it, etc.)
Usually, S/N is measured with the maximum "signal", especially if you are the manufacturer.  ;)    But, it's up to the record company to define their maximum.   I don't know if there is a universal 0dB reference for records.      The RIAA curve complicates things because the "dB" level is going to be measured before EQ (but probably after bandpass filtering).

Vinyl loudness wars existed before digital loudness wars...  It's a question of how loud you can go without mis-tracking and of course that depends on the playback cartridge.   I remember a million years ago, Stereo Review Magazine would test cartridge tracking with the cannon shots from a particular recording of the 1812 Overture. 

Quote
Separately, I'm curious about the 18 kHz figure for frequency response.
I don't know where that came from.  Certainly vinyl is capable  of going above 20kHz.   I believe the highest and lowest frequencies are rolled-off for tracking and to maximize the "signal" with stuff you can hear.   Again, that's up to the record company. 

Quote
Oh, and does anyone know of the type of noise vinyl records exhibit? (or maybe the record plus the preamp?) Is it pink? Brownian? It seems like it's one of them, but turns white past some point.
The worst is impulse noise ("snap", "crackle", and "pop") and I assume there's white noise behind that.   That's all RIAA filtered, which I believe would be between pink and Brownian.    Of course the preamp does add noise.   I've mostly heard hum, perhaps with some RIAA filtered white noise.    It's been a long time since I listened to records, but from what I remember preamp noise was only noticeable when the record wasn't playing...  Record surface noise was the BIG problem.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #118
Coming back to the original topic of this thread, I have some questions.

I've seen the 65 dB figure around and I can definitely agree on that, having looked at the noise of some needledrops. However, how good or bad does it get and within what tolerance of a "max signal" are they referenced? For example, I've seen in the manuals of various tape recorders a lot more standardization with regards to these tolerances. The S/N ratio is calculated at a fixed reference, or explicitly a number of dB above the reference. (they may note their S/N is at reference, or at +3 dB above it, etc.) Additionally, I've seen some figures for third-harmonic distortion for different signal levels relative to a reference.

Analog clipping is the equivalent of digital FS.

The LP equivalent of clipping is trackability. Please see the 2 attachments for representative graphs of trackability.

Quote
Are there any as-detailed looks at specs on lathe cutting or are there too many more variables? (cutting, THEN the performance of the playing needle!) If there a reference amplitude for disc cutting? Any figures for distortion as it relates to level above any such reference? And then signal-to-noise ratio relative to that reference?

Lathes have been able to cut untrackable records for many generations of lathes. Indeed, one of the problems of cutting LPs is cutting something that can be tracked by the customer's gear. If your target market is the mass market including people with lower incomes, the boss makes you be very conservative about making LPs that track on cheap players with say, crystal cartridges.

Quote
Separately, I'm curious about the 18 kHz figure for frequency response. With what reference is this chosen to be the max as well? Some amount of distortion or some amount of roll off? I've seen spectrograms which show distortion products creating 'false' high-frequency content, but does anyone know of any more precise tests? Any pure HF tones cut into vinyl? Maybe sweep up to 20 kHz and see how the original fares vs the added distortion?

If you can track it, you can equalize it to be flat. In modern times there is not too much reason to be worried about raw frequency response as long as the dynamic range is available.

Quote
Oh, and does anyone know of the type of noise vinyl records exhibit? (or maybe the record plus the preamp?) Is it pink? Brownian? It seems like it's one of them, but turns white past some point.

The noise floor has several contributors that can vary all in their selves like turntable rumble. You could occasionally "hear" noises in the recording studio and recording gear on LPs, but on CDs it is they who set the effective noise floor.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #119
Usually, S/N is measured with the maximum "signal", especially if you are the manufacturer.  ;)    But, it's up to the record company to define their maximum.   I don't know if there is a universal 0dB reference for records.      The RIAA curve complicates things because the "dB" level is going to be measured before EQ (but probably after bandpass filtering).
There is for tape... it would be nice if records had one. So, in reference to what has any maximum been used? Just for cutting the records?

Also, I see some test LPs use terms of "+0dB" or some number above that; for example, Hifi news test lp has a track called "Bias setting: 300Hz tone, both channels +18dB". 18 dB above what exactly?
Vinyl loudness wars existed before digital loudness wars...  It's a question of how loud you can go without mis-tracking and of course that depends on the playback cartridge.   I remember a million years ago, Stereo Review Magazine would test cartridge tracking with the cannon shots from a particular recording of the 1812 Overture. 
Ohoho! Was it the Telarc recording?
I don't know where that came from.  Certainly vinyl is capable  of going above 20kHz.   I believe the highest and lowest frequencies are rolled-off for tracking and to maximize the "signal" with stuff you can hear.   Again, that's up to the record company. 
Definitely, but I'm only curious about the relationship between what is still in the original input signal into the lathe and what distortion gets added. Even if it is rolled off, a bunch of HF distortion is added, it seems.
The worst is impulse noise ("snap", "crackle", and "pop") and I assume there's white noise behind that.   That's all RIAA filtered, which I believe would be between pink and Brownian.    Of course the preamp does add noise.   I've mostly heard hum, perhaps with some RIAA filtered white noise.    It's been a long time since I listened to records, but from what I remember preamp noise was only noticeable when the record wasn't playing...  Record surface noise was the BIG problem.
Only the "noise floor" I meant, sorry for being unspecific! But yeah, it seems they tend to make the noise quite "pinkish".
Analog clipping is the equivalent of digital FS.

The LP equivalent of clipping is trackability. Please see the 2 attachments for representative graphs of trackability.
Lathes have been able to cut untrackable records for many generations of lathes. Indeed, one of the problems of cutting LPs is cutting something that can be tracked by the customer's gear. If your target market is the mass market including people with lower incomes, the boss makes you be very conservative about making LPs that track on cheap players with say, crystal cartridges.
Those graphs are telling of what range of frequencies they can track, but are they any use towards getting some solid figures in terms of dB for Hz? (is cm/s ~ dB? or maybe log(cm/s)?) It seems like a cartridge's frequency response would be better-suited. And even then, it varies for what catridge is picking it up.

Hm, these seem to be a bit difficult to factor into a conclusive measure of the "performance" of the format. Having the system limited in all these seperate ways.

If only this were easier! The "tape varies from around 10-13 'bits'" was so nice and easy to verify, too nice... dang LPs.

Perhaps just using needledrops and analyzing an unmodulated groove is all we can do here.

On that note, do you or anybody here know how the frequency response of cartridges is determined? I imagine doing trackability would give too variable resuts and using any sort of test LP wouldn't help either in accuracy. Do they have some special testbed where they can test this, to wiggle the needle around accurately?

If you can track it, you can equalize it to be flat. In modern times there is not too much reason to be worried about raw frequency response as long as the dynamic range is available.
I suppose not, but getting a similar result of how distortion rises with frequncy on an LP would be useful. Maybe a the cutting head from a lathe has such a spec?

Thank you for your responses.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #120
There is for tape... it would be nice if records had one. So, in reference to what has any maximum been used? Just for cutting the records?

Also, I see some test LPs use terms of "+0dB" or some number above that; for example, Hifi news test lp has a track called "Bias setting: 300Hz tone, both channels +18dB". 18 dB above what exactly?
...

The convention is that "0dB" is a maximum stylus lateral velocity of 5 cm/sec at a frequency of 1 KHz.

http://www.audiomisc.co.uk/HFN/LP1/KeepInContact.html
... and the following 4 pages. 
Regards,
   Don Hills
"People hear what they see." - Doris Day

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #121
Quote
On that note, do you or anybody here know how the frequency response of cartridges is determined?
IIRC - In the olden days, the magazines would tell you what test record they used.   But, I don't recall the manufacturer's publishing the test record they use.

Quote
If only this were easier! The "tape varies from around 10-13 'bits'" was so nice and easy to verify, too nice... dang LPs.
Tape noise is more constant, but there are still lots of variables...   Different tape speeds, track width (with pro & multi-track recorders), different tape formulations, different machine calibration, etc.    The tape heads (and overall machine) will affect performance too.




Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #122
Maybe a the cutting head from a lathe has such a spec?

One of the most comprehensive papers on a LP cutting system is: "The Westrex 3D StereoDisk System" authors: Nelson, Carl S.; Stafford, Jerome W. Affiliations: Westrex Division of Litton Industries, Hollywood, CA ; Litton Data Systems Division, Canoga Park, CA
JAES Volume 12 Issue 3 pp. 178-185; July 1964
Publication Date:July 1, 1964

Unfortunately, this paper says nothing about the nonlinear distortion of the cutter head itself. My impression is that performance of the cut LP during playback was far more interesting, but out of the scope of the paper. Attached is the frequency response of the cutter with its linearizing feedback system engaged.

I  was struck by thoughts about the futility of Luddite audiophiles who eschew feedback, but worship LPs cut with this device which obviously used feedback to great advantage.


Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #123
What sample rate and  bit depth , dithered, would create a digital recording with the same bandwidth and signal to noise ratio as vinyl

The attachment shows how bit depth can relate to noise and distortion. Based on what I've seen in the way of measurements, the LP is in the range of 11 bits.

Re: Vinyl is equivalent to which digital bit-depth and sampling rate?

Reply #124
The convention is that "0dB" is a maximum stylus lateral velocity of 5 cm/sec at a frequency of 1 KHz.

http://www.audiomisc.co.uk/HFN/LP1/KeepInContact.html
... and the following 4 pages. 
Extremely useful article, thank you!! Helped answer most of my questions. Also, fitting username!

Quote
On that note, do you or anybody here know how the frequency response of cartridges is determined?
IIRC - In the olden days, the magazines would tell you what test record they used.   But, I don't recall the manufacturer's publishing the test record they use.
Looking into it, it seems to be true from some cursory searches. Never thought it went that way!

Tape noise is more constant, but there are still lots of variables...   Different tape speeds, track width (with pro & multi-track recorders), different tape formulations, different machine calibration, etc.    The tape heads (and overall machine) will affect performance too.
But still, thankfully, quite standardized on each. I see many spec sheets include tables with different weightings, different references, and giving an associated Third Harmonic Distortion.

One of the most comprehensive papers on a LP cutting system is: "The Westrex 3D StereoDisk System" authors: Nelson, Carl S.; Stafford, Jerome W. Affiliations: Westrex Division of Litton Industries, Hollywood, CA ; Litton Data Systems Division, Canoga Park, CA
JAES Volume 12 Issue 3 pp. 178-185; July 1964
Publication Date:July 1, 1964

Unfortunately, this paper says nothing about the nonlinear distortion of the cutter head itself. My impression is that performance of the cut LP during playback was far more interesting, but out of the scope of the paper. Attached is the frequency response of the cutter with its linearizing feedback system engaged.

I  was struck by thoughts about the futility of Luddite audiophiles who eschew feedback, but worship LPs cut with this device which obviously used feedback to great advantage.
Shame about the lack of distortion info, but that curve is interesting! Probably explains some parts of "the sound" associated, at least, with records cut with this lathe. Unless any "newer" ones have gotten flatter.