A few months ago a friend asked me about what I would recommend for archiving his vinyl LP's.
My suggestion was a CCD camera with a microscope lens using a linear tracking mechanism to follow the groove. I postulated that the if the microscope were to read the groove at say a 30 degree angle, with a light source at an opposite 30 degree angle, that their would be sufficient shadow information to adequately understand the contour of the groove. Recording at 30 FPS would be optimal, but less than real time captures would work also. The key is to get accurate pictures of the groove. A stepping motor would be used to rotate the disk. Stepping motors would also position the camera on top of the groove.
Once a good optical rendition of the disk was made, the audio could be rendered at appropriate sample rate (limited by the resolution of the recorded images). Don’t like the RIAA eq? Re-render it with an improved algorithm. Suppose there are likely a few patents to be had here... I just don’t seem to be that entrepreneurial these days.
"Wired" Ran this article today about a clever person who uses their document scanner to read their LP's. Although anything but archive quality, it’s an interesting read nonetheless.
http://www.wired.com/news/digiwood/0,1412,57769,00.html
My suggestion was a CCD camera with a microscope lens using a linear tracking mechanism to follow the groove. I postulated that the if the microscope were to read the groove at say a 30 degree angle, with a light source at an opposite 30 degree angle, that their would be sufficient shadow information to adequately understand the contour of the groove. Recording at 30 FPS would be optimal, but less than real time captures would work also. The key is to get accurate pictures of the groove. A stepping motor would be used to rotate the disk. Stepping motors would also position the camera on top of the groove.
Once a good optical rendition of the disk was made, the audio could be rendered at appropriate sample rate (limited by the resolution of the recorded images). Don’t like the RIAA eq? Re-render it with an improved algorithm. Suppose there are likely a few patents to be had here... I just don’t seem to be that entrepreneurial these days.
"Wired" Ran this article today about a clever person who uses their document scanner to read their LP's. Although anything but archive quality, it’s an interesting read nonetheless.
http://www.wired.com/news/digiwood/0,1412,57769,00.html
hehehe
why not just scan the whole thing and then run a suitable algorithm ?
that means even damaged and broken lps can come back to life !
infinite re-playing without deterioration...
🙂
s
why not just scan the whole thing and then run a suitable algorithm ?
that means even damaged and broken lps can come back to life !
infinite re-playing without deterioration...
🙂
s
I think that’s what the guy in the article did.
His first issue was that its tough to find a scanner large enough for a whole LP. He solved this but making 4, quarter scans and splicing them together.
I know if I tried this it would inevitably sound like a record that had been sawed into 4ths and then stock back together with a hot poker.
He solved the issue of following the groove. An interesting issue for sapling: LP's are recorded Constant Angular Velocity (CAV). As you follow the groove towards the center, the linear velocity will increase.
The big issue he had with his technique was limited fidelity. Someone in the article postulates that this is because the perpendicular angle, and flat lighting of the image, fails to resolve the intricacies of the modulation of a stereo recording within the groove.
I figured that with one of those Intel USB microscopes (available surplus NIB for about $50 on the web), the groove could be followed (although I haven’t done ANY calculations) and recorded. My gimmick is to set the light source and camera at angles to accentuate the contour of the groove’s wall.
If its a 640x480 CCD, I figure I want about the center third of the image to be the groove to analyzed. Lets say that gives us 210 pixels of resolution across the groove. For arguments case well forget about things like sub pixel interpolation and just guess that we can measure to variations of each wall with roughly 50 pixels of resolution. That should yield more dynamic then enough dynamic range. The distance on the wall of the groove between which samples are taken can be adjusted for sample rate and frequency response.
Now I wonder if the time to sample and record this pre-processed image is going to be faster or slower that a real time playback of an LP?
Of course these days I have enough trouble finding time to keep the cats fed and my laundry done. Suppose I’ll never actually do it.
His first issue was that its tough to find a scanner large enough for a whole LP. He solved this but making 4, quarter scans and splicing them together.
I know if I tried this it would inevitably sound like a record that had been sawed into 4ths and then stock back together with a hot poker.
He solved the issue of following the groove. An interesting issue for sapling: LP's are recorded Constant Angular Velocity (CAV). As you follow the groove towards the center, the linear velocity will increase.
The big issue he had with his technique was limited fidelity. Someone in the article postulates that this is because the perpendicular angle, and flat lighting of the image, fails to resolve the intricacies of the modulation of a stereo recording within the groove.
I figured that with one of those Intel USB microscopes (available surplus NIB for about $50 on the web), the groove could be followed (although I haven’t done ANY calculations) and recorded. My gimmick is to set the light source and camera at angles to accentuate the contour of the groove’s wall.
If its a 640x480 CCD, I figure I want about the center third of the image to be the groove to analyzed. Lets say that gives us 210 pixels of resolution across the groove. For arguments case well forget about things like sub pixel interpolation and just guess that we can measure to variations of each wall with roughly 50 pixels of resolution. That should yield more dynamic then enough dynamic range. The distance on the wall of the groove between which samples are taken can be adjusted for sample rate and frequency response.
Now I wonder if the time to sample and record this pre-processed image is going to be faster or slower that a real time playback of an LP?
Of course these days I have enough trouble finding time to keep the cats fed and my laundry done. Suppose I’ll never actually do it.
I hadn't read the article/website before posting earlier.
I think a higher resolution scanner would be essential otherwise the microscope idea sounds intriguing too.
I've seen microgroove pictures taken before but I don't know how these were taken, i.e. under uniform light, angled light or what ?
Some experimenting would be in order there.
If the depth of the groove is not constant (I'm sorry I don't know enough about this) then a vertically taken picture should be enough to recover all the information.
If this is not correct (which is what I suspect) then a holographic (or maybe just stereoscopic ?) image would be necessary. If it were possible to create such an image through a different scanning technique then the whole process might be very cheap.
The stepping motor + USB microscope sounds very cheap too but a little challenging constructionally. However, I'm not sure if I understood completely the way the microscope and the light source would be used.
My guess is that to obtain the information recorded in the varying grove depth, at least two microscopes need to be used simultaneously.
s
I think a higher resolution scanner would be essential otherwise the microscope idea sounds intriguing too.
I've seen microgroove pictures taken before but I don't know how these were taken, i.e. under uniform light, angled light or what ?
Some experimenting would be in order there.
If the depth of the groove is not constant (I'm sorry I don't know enough about this) then a vertically taken picture should be enough to recover all the information.
If this is not correct (which is what I suspect) then a holographic (or maybe just stereoscopic ?) image would be necessary. If it were possible to create such an image through a different scanning technique then the whole process might be very cheap.
The stepping motor + USB microscope sounds very cheap too but a little challenging constructionally. However, I'm not sure if I understood completely the way the microscope and the light source would be used.
My guess is that to obtain the information recorded in the varying grove depth, at least two microscopes need to be used simultaneously.
s
You have to read off both sides of the track, so I think the idea
of having one microscope at an angle is questionable. However,
as an alternative to using two microscopes, as stelios suggested,
I was thinking of a slightly different approach. Use one microscope
at 90 deg. to the surface, but use two lightsources each at a
suitable angle, but on opposite sides of the track. My first idea
was to switch these on/off alternately, which would cause some
other problems. Either you use a stepper-motor and read a piece
at a time under two different lighting conditions, or you switch
the light sources so quickly that you can rotate the record at
a constant speed and stich the infor together in software. While
writing this I got another idea, however. Maybe it won't work
well since the records are usually black, but instead of switching
the two light sources on/off alternately, they could have different
colour so the reflected light can be separated in software as
coming from either light source.
of having one microscope at an angle is questionable. However,
as an alternative to using two microscopes, as stelios suggested,
I was thinking of a slightly different approach. Use one microscope
at 90 deg. to the surface, but use two lightsources each at a
suitable angle, but on opposite sides of the track. My first idea
was to switch these on/off alternately, which would cause some
other problems. Either you use a stepper-motor and read a piece
at a time under two different lighting conditions, or you switch
the light sources so quickly that you can rotate the record at
a constant speed and stich the infor together in software. While
writing this I got another idea, however. Maybe it won't work
well since the records are usually black, but instead of switching
the two light sources on/off alternately, they could have different
colour so the reflected light can be separated in software as
coming from either light source.
Would an optical (visible light) sytem have enough resolution to read an LP groove ? At best you could reach 1/4 wavelength of the incident light - no amount of magnification will get beyond this.
Dave
Dave
DRC said:
For a price of $10000+ I presume these beasts work
http://elpj.com/frameset.html
which would suggest that the wavelength is not a problem.
Whether the microscope method would have sufficient resolution
is a different issue, though.
- Status
- Not open for further replies.