Project: Homemade optical record player.

[Mr.TalkingMachine]

First of all, a bit of background.

I have been interested in gramophone records since I was a very young teen, and a (perhaps mediocre) EE student since three years ago.

About a month ago I came across this item. An exceedingly rare French Pathé record made around the year 1905 (pictures attached). These records were made of a cement core and a coating of phonograph cylinder wax, in which the grooves were stamped. These were made for a very short period of time, and as you can see, as they age, different thermal coefficients cause the wax layer on these records to crack all over and crumble. As if it weren't enough, just like wax cylinders, they can be attacked by mold (these disks really don't want to exist).
Personally I've never heard of anyone who has ever found a complete, playable one.



Bits of this record continue to fall from it, as whatever is left of the ancient recording turns into shambles; however, there's still a groove on it, and looking at it through a lens I can see there seems to be sound still recorded in it. There is a surviving phonograph cylinder of the same song and performer you can listen, but it's possible it's two different recordings.

Jocelyn's Berceuse by Mary Boyer, c.1902 on cylinder:
http://cylinders.library.ucsb.edu/search.php?queryType=@attr+1=1020&num=1&start=1&query=cylinder14802

Mechanical playback is off the table, as it would likely destroy both the record and the stylus, but I believe a special optical player would be more suitable for archiving this recording to a digital format.

There are optical players in the market, but they are very expensive, and they are most likely incompatible with this unusual record for various reasons which go as follows:

-The recording speed is between 90 and 100 RPM, otherwise unspecified, which is non standard.
-The grooves are vertical cut, this means the stylus vibrates up and down, instead of side to side like in normal mono records.
-The record is centre start, it plays from the label towards the rim.
-The record is badly decayed, with a dull surface, lots of cracks and chunks of it missing.

What I seek to do is to build my own optical record archiving machine, capable of producing an audio transfer from the record to a digital format, whether or not it's capable of playing it in real time is not important. The tech developed will be of value, even if this record turns out to be unrecoverable.
As I have low resources, use as much recycled and low cost parts as possible.

The mechanical matters of spinning the record, and tracking the read head over it are secondary goals, my main goal right now is to work on the read head, for which I intend to modify or otherwise re purpose read heads from CD drives.

What concerns me the most is how to recover the audio from the light reflected back from the record, and if it's possible to reverse engineer the existing sensor on a CD drive carriage for this purpose. If possible, I seek help researching these sensors.
Here https://patents.google.com/patent/US3992593A/en is the Google Patents entry for the original optical record player, dating back to the 1970's.

One thing worth noting is the feedback system used for tracking the head, specified in the patent. I'll probably have to come up with a different method for this due to the missing chunks of the record. I think that I could use the read head to first count the grooves and calculate the groove density of the record to calculate the tracking speed afterwards.

[Kjelt]

Good luck, big project!
I was just thinking if it would not be easier to use some sort of high resolution z-axis measurement device, to mechanically probe the depth of the groove. Usb camera to track the groove and a stepper to turn the disc. Anyway interesting project.

[Mr.TalkingMachine]

I actually read once about some curator who did just that to a bunch of Edison cylinders of Native Americans, only he probed the depth using a different method which I cannot remember.
He usually used photography for run off the mill 78 RPM records, which is easier to do, as the grooves are easy to see and they deflect side to side.

The grooves on this are very thin and hard to see, thankfully there's the cylinder transfer of this song.

[chris_leyson]

CD player optics won't be much good for this application as they are designed for tracking and focus for 1.6um spaced tracks. Also, the polycarbonate layer between the aluminium layer and the surface of the disc is also part of the optics. http://leung.uwaterloo.ca/MNS/102/Lectures%202007/Lect%20Mat/physics%20of%20cd%20pe930103.pdf. I think the 2D IRENE scanning approach used by Carl Haber is your best bet, the optics is a lot simpler http://irene.lbl.gov/

[mikeselectricstuff]

I'd approach it by letting software do the hard work. Figure out a lighting system, lighting at an angle  that allows a camera to capture the groove information, from a camera viewing the rotating cylinder ( multiple tracks at a time)  and use something like OpenCV to process it.

[Mr.TalkingMachine]

Interesting suggestion, I'll look into OpenCV, although it's gonna take me a while to learn enough of this.

I am not a software person, my sauce, so far, is analog electronics.

[richnormand]

Look at this link for several options and ideas.

https://en.wikipedia.org/wiki/Laser_turntable

Any non-destructive ways to take a molding of the record. That way the molded relief would have all the same colour.

I look forward to your progress.

[Kleinstein]

If this is for looking at a few old disks, one may not need to have the system operate in real time. This can ease things up,a s the computer could use more time looking at the pictures. So it won't be a direct player but more like a format converter:  groovy part in and wave file out.

[mikeselectricstuff]

Do you have rough dimensions of the groove ?

[edavid]

I think the 2D IRENE scanning approach used by Carl Haber is your best bet, the optics is a lot simpler http://irene.lbl.gov/

The vertical grooves would require the 3D scanner.

OP, if you look at Carl Haber's materials, you'll see this is not a garage project.

[amyk]

Do you have rough dimensions of the groove ?

That's what I was going to ask too. It's possible this early record has far bigger tracks than the "standard" ones later on, and using a (suitably low-powered) laser and photodiode might be enough to recover significant amounts of the data.

[paulca]

CD laser reader would be geared towards HIGH or LOW.  You need to read analogue levels.  The groove is not on and off, but an actual analogue waveform.  If you consider input to an ADC you need to distinguish between possibly 1024 different positions of the groove side, or more.

Assuming it's mono that at least simplifys things.

If you hooked up the CD laser sensor and read it's analogue outputs on oscilloscope as it reflects off the surface it would pass or fail that approach.  Read one groove, then worry about groove tracking.  There might be analogue values enough to determine the main gist of the groove shape with the right lighting angle.

I don't expect the quality will be very high, most of which will be from the condition of the groove and even when it was brand new it can't have been that good, but I suppose the least amount of noise you can add and the most signal you can read the better.

Most other "light" or "sound" based approaches will suffer from focus, you need a light/sound sensor that can be focused on a very, very small area.

[dmills]

Why bother spinning the disk?

Laser of reasonable coherence length (He-Ne or SLM diode module?), beam splitter, some assorted optics, a sand table in a low vibration environment and some suitably slow photographic film.

Capture the interference pattern in classic hologram style, then work from the fringes.

You might well be able to find a holographer who thinks this is a cool project.

Regards, Dan.

[Mr.TalkingMachine]

Do you have rough dimensions of the groove ?

I do not, the grooves are very fine and also worn, guess I'll try to take a magnified picture of the surface near a ruler and work from there.

[Mr.TalkingMachine]

I took these pictures, the scale is in millimetres.

For comparison, a Victor record from 1905.



Note that the groove density is about the same, but the Pathé grooves are nowhere near as deep.

[phil from seattle]

I think it would be pretty easy to build a machine to read those discs.  I'm sure SW could be designed to take a high res photo and convert it but a simple turntable with a laser pickup is pretty easy to build.  I suspect even an open loop design would work.

[oldway]

Why use an electronic solution ? The problem is that it can't be played because the surface is too fragil....Solution is simple: use the same process that is been used to make vinyl copy from a fragile lacquer disk ....Make a matrix , see from step 4 onwards.....



You can then play the copy without damaging the original.

[phil from seattle]

The video presumes that the lacquer master is not unique - it can be recut from the master tapes (or what ever storage) in the event of a problem. I don't see how that fragile disc could be used as a master without damaging it. A contactless approach is really the only solution that preserves the original.

[Twoflower]

I would see if a photographic approach will do. But for the Pathé it will take much time to get enough contrast out of the track to reveal enough of the variation. For the Victor that seems possible as you already see the 'sound' in the provided picture.

Taking many well done macro pictures with light shining into the grove wile turning the disk step by step will probably be a nice starting point. Stitching should be possible by using panorama SW. But the final step to turn it to an audio file will be interesting. But not impossible as some used a scanner to play a LP (extremely bad quality, but it's more a prove of concept). This SW might need the information of the rotation point, so include that centre part of the disk in the scanning process.

[oldway]

The video presumes that the lacquer master is not unique - it can be recut from the master tapes (or what ever storage) in the event of a problem. I don't see how that fragile disc could be used as a master without damaging it. A contactless approach is really the only solution that preserves the original.

Are you a specialist in the manufacture of vinyls?
How could applying a single layer of silver on the disc damage it?
The adhesion is very low, one must even add tin chloride to increase it.
How could you know the exact conditions of the disc when you have not had it in your hands?

I think you make assertions a bit too quickly without having the needed informations to do them.

[Mr.TalkingMachine]

How could applying a single layer of silver on the disc damage it?
The adhesion is very low, one must even add tin chloride to increase it.
How could you know the exact conditions of the disc when you have not had it in your hands?

I have already stated the disk is so fragile it's basically self destructive. Any sort of contact causes pieces of it to fall off.

Trying to take a stamper would likely cause all the wax to come off at once.

[oldway]

How could applying a single layer of silver on the disc damage it?
The adhesion is very low, one must even add tin chloride to increase it.
How could you know the exact conditions of the disc when you have not had it in your hands?

I have already stated the disk is so fragile it's basically self destructive. Any sort of contact causes pieces of it to fall off.

Trying to take a stamper would likely cause all the wax to come off at once.

That fragile ? Ok, I understand, be careful not to catch the flu ... a sneeze and that's the end of your project. 

And what about an ultrasonic beam? Should it be possible to make a very tiny ultrasonic beam to read the track by measuring the distance?

[mikeselectricstuff]

There's always the x-ray option... https://hackaday.com/2018/01/02/60s-comedy-rescued-from-nigerian-basement/#comment-4285065

[paulca]

I wonder if an MPEG motion vector file would give you anything if you simply videoed the tracks under bright light and high zoom.  It will give you a signal for every pixel that moves, it's motion direction and displacement.   (EDIT:  actually it will give you the vector for a "pel" which is usually and 8x8 group)

An MPEG motion vector file can be got in textual form from various encoders or motion stabilisation software.  They are also often used for multi-pass encoding where the average motion of the image is recorded to work out how much bit rate to devote to best achieve the average bit rate of the encoding.  No point wasting bandwidth when the video is mostly static.  Save it for the fast action scenes.

[LaserSteve]

Used HP Laser Interferometer to measure distance, linear bearing slide driven by a linear motor such as Parker Daedal,  and a 32 KHz quartz crystal  AFM probe to sense contact.  The AFM probe can see single atoms in its normal role.  I have a system I built at work that hits 150 nm resolution,  so a record at a few microns is definitely doable. That was done with used gear, graduate student labor, and one hell of a skilled programmer.  Other then what I just said, I cant give you schematics for a variety of legal reasons, not to mention the device I designed is set up for measuring the average top surface  soft fuzzzy non-woven stuff, not probing an  analog  record.

Your really in the ream of a AFM, an atomic force microscope...  You dont need the sub nanometer resolution, but it will give you the lowest, most controlled force possible. You can also use optical detection of the focus, that gear is off the shelf but expensive.

Before you go to the problem of locating a used HP interferometer for the motion control, talk to the US National Archives..  I bet they know who can read your disk.

If I had time, say four  months, access to a really good machine shop, and 10-15,000$ of used instruments and materials, with about 5-7K$ of that going to a very skilled assembly langage  programmer....
Well maybe... But I dont have time and no intention of leaving my present employer.. That is the I'm retired or wealthy and its a labor of love cost, doing it commercially would add another 20-40K$ in labor...

So try the US National archives and maybe NIST...  Nist probably has some very nice tools for metrology over large surfaces...

A friend of mine did get the two axis  HP interferometer to USB  down to about 200$ for the interface... So its doable...

Steve