HydrogenAudio

CD-R and Audio Hardware => Vinyl => Topic started by: Apesbrain on 2012-12-21 20:16:46

Title: At long last, copy your digital collection to vinyl!
Post by: Apesbrain on 2012-12-21 20:16:46
http://www.engadget.com/2012/12/21/3d-printed-record/ (http://www.engadget.com/2012/12/21/3d-printed-record/)

Enjoy the video!
Title: At long last, copy your digital collection to vinyl!
Post by: Sunhillow on 2012-12-21 20:25:05
This brings the analog warmth back to digital sources
Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-21 20:53:13
I wouldn't advise this! The stylus isn't playing a smooth waveform, it is attempting to play a staircase textured groove wall. The constant bashing of the needle between steps is analogous (pardon the pun) to applying an improperly light, downward tracking force with a normal record, also very bad. Besides the distortion, it wears down the needle very quickly with all the side to side groove bashing going on!
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-21 21:31:42
I wouldn't advise this! The stylus isn't playing a smooth waveform, it is attempting to play a staircase textured groove wall.


If a modern needle is 0.003 inches wide, and is (often) elliptical and around 0.012 inches long and the resolution of the 3D printer used was 600dpi (0.002) I don't see what the concern is.  The stylus is over five times as long as the coarsest grain on the sidewall.  In actuality 3D printer's don't shit out perfect little 0.001666666" cubes, and so the actual coarseness would be even less, no?

Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-21 22:26:09
What matters is not the dimensions of the needle, but rather the dimensions of its contact area to the steps of the staircase, er , groove wall. [or is that what your specs were?] The severe distortion we hear in this 3D printed LP, linked to, I assume isn't due to the limitations of the encoding process, nor the limitations of the 3D printer's inability to make perfect cubes, but rather in the needle playback; it is severly, for lack of a better word, "mistracking". It certainly sounds just like conventional mistracking due to, say, an inadequate tracking force. Mistracking needles kill the record surface with as little as one play and slowly kill themselves, or at least wear down pre-maturely.
Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2012-12-22 02:42:28
The "stairstepping" of the printed groove walls is what causes the "fire siren" background sound. It's noticeable, but not as bad as you imply. The actual music distortion is due to two causes: Lack of RIAA pre-equalisation, and lack of printer resolution (the "wiggles" in the groove walls for the higher frequencies are smaller than the printer resolution.)

The point is not that the quality is poor at the moment. Printers (and thus the quality) will improve, and their costs will drop.
Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-22 03:32:59
The "stairstepping" of the printed groove walls is what causes the "fire siren" background sound. It's noticeable, but not as bad as you imply.

Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.

To the needle, as it plays these impossible to properly track squares, the signal seems to be a signal which alternates between square waves to triangular waves, twice per 1.8 second rotation: square, triangle, square, triangle and then back to square. But since it largely just smashes through ignoring them, all we hear is the fire siren whine rendition. [and yes, some music too]

Quote
Lack of RIAA pre-equalisation
If that's true, then we also aren't hearing this LP played back with RIAA de-empasis either, via a phono preamp, because the music we hear does have a fairly normal bass content and is not thin and squeaky. [The "LP" play time per side would be much smaller too.]

I'm not claiming this method might not eventually be just as good if not better than conventional cutterheads and lathes as 3D printers improve (if there was any reason to continue making LPs), I'm just commenting on how terrible this version sounds, at least to my ears.
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-22 03:50:18
Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Sure they do.  Just at a much smaller scale.  My point was that the needle doesn't care once you get past a certain point.  I'm not sure 600dpi is that point, but it isn't far off.

To the needle, as it plays these impossible to properly track squares, the signal seems to be a signal which alternates between square waves to triangular waves, twice per 1.8 second rotation: square, triangle, square, triangle and then back to square. But since it largely just smashes through ignoring them, all we hear is the fire siren whine rendition.


But it isn't a series of squares and triangles to the needle.  The needle is too large to "feel" the inner corner (triangle) and won't ride the flats (square) as the leading edge of the elliptical stylus is limited in its range of motion by the trailing edge is limited by the "side (and vise versa).
Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-22 04:16:36
Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Sure they do.  Just at a much smaller scale.

No, conventional analog records have no staircase structure to their groove walls, at all, hence no staircase structure noise. They do have lots of other noises, however, including groove wall noise which, at times, is modulated per rotation.
Title: At long last, copy your digital collection to vinyl!
Post by: Nessuno on 2012-12-22 09:06:23
The printing process is actually a downsampling and then the needle act as an integrator: a "two stage mechanical DA converter"! Now, if only there would be a couple of triodes in the printer cartridge driver...

C'mon, boys! 
Title: At long last, copy your digital collection to vinyl!
Post by: Glenn Gundlach on 2012-12-22 09:46:36
I wouldn't advise this! The stylus isn't playing a smooth waveform, it is attempting to play a staircase textured groove wall. The constant bashing of the needle between steps is analogous (pardon the pun) to applying an improperly light, downward tracking force with a normal record, also very bad. Besides the distortion, it wears down the needle very quickly with all the side to side groove bashing going on!


Have you ever looked at the analog output waveforms of a properly operating digital system? There are no little steps as they get completely cleaned up in the reconstruction filters. If there ARE steps it was either planned or something needs repair. Telarc LP discs in the late '70s and '80s were made from digital masters are were regarded as some of the best LPs you could get - if you like that sort of thing. Study the physics of the LP and you'll come to the conclusion it's a monstrosity of a mess and amazing that it works as well as it does.

Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-22 10:47:13
Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Sure they do.  Just at a much smaller scale.

No, conventional analog records have no staircase structure to their groove walls, at all, hence no staircase structure noise. They do have lots of other noises, however, including groove wall noise which, at times, is modulated per rotation.

You're not looking at a small enough scale.

I wouldn't advise this! The stylus isn't playing a smooth waveform, it is attempting to play a staircase textured groove wall. The constant bashing of the needle between steps is analogous (pardon the pun) to applying an improperly light, downward tracking force with a normal record, also very bad. Besides the distortion, it wears down the needle very quickly with all the side to side groove bashing going on!


Have you ever looked at the analog output waveforms of a properly operating digital system? There are no little steps as they get completely cleaned up in the reconstruction filters.

Glenn, mzil is talking about the physically dot-printed walls, nothing to do with what you're talking about.
Title: At long last, copy your digital collection to vinyl!
Post by: Glenn Gundlach on 2012-12-22 18:41:51
Glenn, mzil is talking about the physically dot-printed walls, nothing to do with what you're talking about.


That's what I get for being a know-it-all and not looking at the link. Sorry 'bout that.

Title: At long last, copy your digital collection to vinyl!
Post by: bryant on 2012-12-22 23:05:36
This is very cool! I recently had sort of the reverse idea: storing digital audio on [pressed] analog LPs. We used to store digital data on analog cassettes (I’m dating myself here), but the technology was extremely primitive and the system I used was only about 1 kbps, although I have read that people got up to 4 kbps with pretty cheap mono machines. Magazines would sometimes publish programs on thin plastic records that you tore out (although those were usually 300 bps, IIRC).

By taking advantage of today’s much more advanced methods (QAM256, Reed-Solomon ECC, etc.) and the LP’s greater bandwidth (20-25 KHz times two channels) it might very well be possible to get 128 kbps, which would provide usable audio quality with Opus or AAC encoding. Of course you would want a little decoder box with an S/PDIF output, but the prototype could use a PC for playback. All the fun of playing LPs without the clicks and pops! (The only LP artifact you’d still be left with is wow, and you could get rid of that if you were willing to put up with some latency).
Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2012-12-23 00:33:01
The "stairstepping" of the printed groove walls is what causes the "fire siren" background sound. It's noticeable, but not as bad as you imply.

Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Of course it's not "listenable" compared to a traditional LP. My point was that even at 600 dpi, it works as a proof of concept. Do you remember the grainy image display on early PCs? 640x480x16 colours? That's about where this "proof of concept" is now. And look how far display image resolution has come...

...
I'm not claiming this method might not eventually be just as good if not better than conventional cutterheads and lathes as 3D printers improve (if there was any reason to continue making LPs), I'm just commenting on how terrible this version sounds, at least to my ears.


Like I said, "The point is not that the quality is poor at the moment. Printers (and thus the quality) will improve, and their costs will drop."

One of the barriers to LP production is the minimum quantity. Like pressing CDs, there's a setup cost to be amortised. The setup cost for CDs can be avoided by using CDRs for small runs. If a "printed" LP can be made with similar fidelity to a pressed LP, the setup cost can likewise be avoided. Given a high enough resolution, a printed LP could sound better than a pressed LP because it avoids the significant limitations imposed by the current production method. (Cutter head limitations, especially at HF. Generational loss though the plating / mother / stamper production process. Stamper wear and handling damage.)


Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-24 05:36:58
Considering normal records have absolutley none, I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Sure they do.  Just at a much smaller scale.

No, conventional analog records have no staircase structure to their groove walls, at all, hence no staircase structure noise. They do have lots of other noises, however, including groove wall noise which, at times, is modulated per rotation.

You're not looking at a small enough scale.

OK then, please explain exactly what causes this staircase structure on a conventional analog record groove wall, you speak of.
Title: At long last, copy your digital collection to vinyl!
Post by: pdq on 2012-12-24 13:42:48
Do you remember the grainy image display on early PCs? 640x480x16 colours?

I certainly remember its predecessor, CGA, with 2 bits per pixel. I don't remember its resolution, but it was a lot less than 640x480.
Title: At long last, copy your digital collection to vinyl!
Post by: Arnold B. Krueger on 2012-12-24 13:47:31
This brings the analog warmth back to digital sources


http://www.engadget.com/2012/12/21/3d-printed-record/ (http://www.engadget.com/2012/12/21/3d-printed-record/)

Claimed performance is "11KHz, with a 5 - 6bit resolution".  22 KHz sampling with 12 bit resolution is listenable. This stuff misses the mark by 100% ;-)
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-24 20:10:38

You're not looking at a small enough scale.

OK then, please explain exactly what causes this staircase structure on a conventional analog record groove wall, you speak of.


At a small enough scale the walls of a record groove are quite rough.  Vinyl molecules are not infinitely small.

My point, though, is that this doesn't matter.  It doesn't matter because the stylus is large enough that it can not resolve the "texture" of the individual vinyl molecules.

At some point (and I'd welcome the math showing exactly when) the limited resolution of the 3D printer doesn't matter either, for the same reason.  Based on the numbers I cited above I strongly suspect that the scale of the printer is already fine enough that the stylus can't resolve the "square" and "triangle" shapes you mentioned, but what exactly the cut off is I don't know.

Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2012-12-24 20:25:08
Do you remember the grainy image display on early PCs? 640x480x16 colours?

I certainly remember its predecessor, CGA, with 2 bits per pixel. I don't remember its resolution, but it was a lot less than 640x480.


CGA did have a 640x480, 2 colour (black and white) mode. There was a whole body of work developing dithering patterns to best display images on such displays, and on the 300dpi single-density laser printers of the period.  Again, the point is that 3D printer resolution will improve and the cost will come down. There are many potential uses eagerly waiting for this to happen, so there's plenty of financial incentive. For example, complex and unique lens curvatures (spectacles, contact lenses) can be printed rather than ground.
Title: At long last, copy your digital collection to vinyl!
Post by: lvqcl on 2012-12-24 20:33:08
AFAIR CGA graphics = 320x200 4 colors or 640x200 2 colors
Title: At long last, copy your digital collection to vinyl!
Post by: washu on 2012-12-24 20:40:26
CGA did have a 640x480, 2 colour (black and white) mode.

Actually it was only 640x200 for CGA black and white mode.  The pixels were non square. The foreground could be set to any of the 15 colours, but the background had to be black.  EGA added 640x350 with more colours.  While the PGA could do 640x480 in 256 colours before VGA came out, it was extremely expensive and rare.  VGA was the first wide spread graphics card that could do 640x480 in x86 PCs.
Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2012-12-24 23:11:04
You're right. Memory is the second thing to go with age. I forget what the first thing was...
I should know better, I even have a working PC Portable and 5153 monitor (given to my father many years ago and, sadly, inherited back.)
Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-24 23:58:56

You're not looking at a small enough scale.

OK then, please explain exactly what causes this staircase structure on a conventional analog record groove wall, you speak of.


At a small enough scale the walls of a record groove are quite rough.  Vinyl molecules are not infinitely small.


There's no vinyl record stylus/laser/optical scanner mechanism known to resolve that level of detail you speak of, not by a long shot; it would be like attempting to examine small specs of dust that appear on the surface of rocks found on the slopes of Mt. Everest from a digital camera photograph of the entire mountain, with the mountain itself representing a typical imperfection found on conventional record groove walls. The molecular bumps, which are the specs of dust on Everest, can't be heard nor can they be recorded with any known technology that scans the music on LPs in one pass, just like you can't blow up an image of Mt. Everest to look for dust particles on the rocks.

Even speaking theoretically, these molecular-level, tiny bumps and pits (which have nothing to do with the overwhelming and relatively mountainous bumps and valleys which make up the real groove wall noise that we actually can hear on a conventional record) are scattered about randomly throughout the entire rotation of the record groove, not in a methodical, grid-like staircase structure like on her printed records have, hence they would appear as ultra high frequency, random noise (think "hiss"), not a fire truck siren tone which rises and falls per 1.8 second rotation (like we clearly hear on her printed records).

No staircase structure = no staircase structure noise. Ergo, conventional records have no 1.8 second modulated fire engine siren noise, at any frequency, at any level, not even theoretically at a molecular level.



Quote
Based on the numbers I cited above I strongly suspect that the scale of the printer is already fine enough that the stylus can't resolve the "square" and "triangle" shapes you mentioned, but what exactly the cut off is I don't know.


Although as it was pointed out earlier (by another forum member) they are more like irregular blobs, not precise little cubes, the stylus of her rather junky turntable most definitely can make out the grid structure of her printed records. I guess the dot structure is more sharply  defined across one of the two horizontal axes than the other, hence she refers to the pattern as being one of closely packed horizontal lines:

"yes the repetitive high pitch frequency sweep is actually a byproduct of the printing process. The objet has an array of heads that it uses to print and it moves this array back and forth as it lays down material. At the end of the print the surface is covered with tiny 600dpi parallel lines caused by the space in between the individual heads. Each time the record needle passes perpendicularly over these lines it adds another frequency into the audio output. if you listen closely, you will hear the frequency of this sweep decrease as the needle moves to the inner grooves (because it does not move as fast across the surface of the record)"

Source: http://www.instructables.com/id/3D-Printed...fset=40#DISCUSS (http://www.instructables.com/id/3D-Printed-Record/?&sort=ACTIVE&limit=40&offset=40#DISCUSS)
amandaghassaei (author) says: Dec 20, 2012. 10:27 AM in the discussion section
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-25 00:14:10
There's no vinyl record stylus/laser/optical scanner mechanism known to resolve that level of detail you speak of, not by a long shot; it would be like attempting to examine small specs of dust that appear on the surface of rocks found on the slopes of Mt. Everest from a digital camera photograph of the entire mountain, with the mountain itself representing a typical imperfection found on conventional record groove walls. The molecular bumps, which are the specs of dust on Everest, can't be heard nor can they be recorded with any known technology that scans the music on LPs in one pass, just like you can't blow up an image of Mt. Everest to look for dust particles on the rocks.

Even speaking theoretically, these molecular-level, tiny bumps and pits (which have nothing to do with the overwhelming and relatively mountainous bumps and valleys which make up the real groove wall noise that we actually can hear on a conventional record) are scattered about randomly throughout the entire rotation of the record groove, not in a methodical rising and falling staircase structure like on her printed records have, hence they would appear as ultra high frequency, random noise (think "hiss"), not a fire truck siren tone which rises and falls per 1.8 second rotation (like we clearly hear on her printed records).

No staircase structure = no staircase structure noise. Ergo, conventional records have no 1.8 second modulated fire engine siren noise, at any frequency, at any level, not even theoretically at a molecular level.


You completely appear to have missed the line:

"My point, though, is that this doesn't matter. It doesn't matter because the stylus is large enough that it can not resolve the "texture" of the individual vinyl molecules."

But thanks for telling me what I already said. 

Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2012-12-25 00:33:47
Considering normal records have absolutley none [fire truck siren noise], I would consider it completly unlistenable, but if you deem it "not so bad", so be it.


Sure they do.  Just at a much smaller scale.



My point is that you are dead wrong; they don't have any, not at any scale, none, not even in theory. But feel free to use strawman arguments and change the subject to "but the real question is... blah-blah-blah" as much as you'd like. I don't care nor will address any such red herrings.
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-25 01:10:14
Again, the point is that your line:

To the needle, as it plays these impossible to properly track squares, the signal seems to be a signal which alternates between square waves to triangular waves, twice per 1.8 second rotation:



Is incorrect.  The stylus is too large to see 600dpi squares and triangles (edit: ahh - except for the four times per revolution the stylus is aligned with the axis of the pinter) just as it is too large to see molecules of vinyl.

There's no red herring, there's no strawman.

I never denied their was aliasing caused by the coarseness of the printing...
Title: At long last, copy your digital collection to vinyl!
Post by: Soap on 2012-12-25 02:47:00
Considering normal records have absolutley none [fire truck siren noise], I would consider it completly unlistenable, but if you deem it "not so bad", so be it.



Noticing now your clarification of your original quote, obviously I'm not claiming normal records have any fire truck siren noise.

But speaking of straw men, I'm not sure where I said I deem the distortion of the printed record "not so bad'.
Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2012-12-25 10:59:48
... My point is that you are dead wrong; they don't have any, not at any scale, none, not even in theory. But feel free to use strawman arguments and change the subject to "but the real question is... blah-blah-blah" as much as you'd like. I don't care nor will address any such red herrings.


Wooo. Pot, kettle, black.

Real LPs may not have fire engine noises. They do have noise due to the size of the vinyl molecules. It sets the lower limit of the resolution that can be realised, just as the size of your eardrums sets the lowest limit on the resolution of your hearing. (If your ears were any more sensitive, you'd hear a constant noise from the air molecules bouncing off your eardrums.)

The following page by Jim Lesurf will give you an appreciation of the dimensions involved - what resolution would be required in a 3D printer to reach the resolution of a traditional LP.

http://www.st-andrews.ac.uk/~www_pa/Scots_...rt12/page2.html (http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/iandm/part12/page2.html)
Title: At long last, copy your digital collection to vinyl!
Post by: mccarthyk on 2013-02-09 16:58:19
That is amazing.  I'm sure it's not too difficult to write a program to make an STL file, but getting access to a 3D printer that large and accurate is either hard, or expensive!
Title: At long last, copy your digital collection to vinyl!
Post by: BFG on 2013-02-09 18:09:21
When I first read this topic's title, I thought someone was trolling the board.
Title: At long last, copy your digital collection to vinyl!
Post by: krabapple on 2013-02-09 18:59:57
I wouldn't advise this! The stylus isn't playing a smooth waveform, it is attempting to play a staircase textured groove wall. The constant bashing of the needle between steps is analogous (pardon the pun) to applying an improperly light, downward tracking force with a normal record, also very bad. Besides the distortion, it wears down the needle very quickly with all the side to side groove bashing going on!



I'd like to think you're joking.  But apparently it's not a spoof.
Title: At long last, copy your digital collection to vinyl!
Post by: mzil on 2013-02-09 20:31:08
I'd like to think you're joking.  But apparently it's not a spoof.

Please clarify. I'm not sure I understand what you mean by "it's" not a spoof. Do you mean what I've written is "not a spoof"?
Title: At long last, copy your digital collection to vinyl!
Post by: splice on 2013-02-10 02:02:42
That is amazing.  I'm sure it's not too difficult to write a program to make an STL file, but getting access to a 3D printer that large and accurate is either hard, or expensive!


It's too hard and expensive now. As Mzil says, even state of the art printers are still too crude to print LPs as good as the traditional method.  But it won't be that way for very long. Resolutions are increasing all the time and costs are coming down. Eventually it'll reach a point where the quality meets or exceeds traditional pressing. It'll reach the same point as CDRs have versus pressed CDs - printing will be cheaper than pressing for short runs. For LPs, this is a good thing - it'll become economic to produce a wider catalog of LPs for what is a relatively small market.