Reading Data From A CD, With A Microscope

There was a time when electronic engineering students studied the audio CD, for all its real-world examples of error correction and control systems. There’s something to be found in the system still for young and old though, and thus we were intrigued when we saw [Peter Monta] reading the data from a CD using a microscope.

CDs encode data as so-called pits and lands in a spiral track across a metalised surface, with a transition from pit to land signifying a logic 1 and a missing transition signifying a 0. Reading a section of the raw data is achieved in the first part of his write-up, but in the next installment he goes further into retrieving more data through stitching together microscope pictures and writing some code to retrieve data frames. He’s not quite at the audio playback stage, but he’s planning in the future to spiral-track a full image to rip an entire disc.

There are plenty of CD drives around to read audio the conventional way, but the techniques here still find a use where less ubiquitous media has to be read. In the last decade for example there was an effort to read the BBC Domesday Project from the 1980s, as it became clear that few of the original readers survived in working order.

18 thoughts on “Reading Data From A CD, With A Microscope

  1. I wonder if something similar could be applied to magnetic storage discs (or tapes). How could one obtain an image of magnetic state of a whole media with enough precision to later decode its content by software?
    For disks, maybe by using a head with fine headgap/track width (maybe HDD one?) and spinning the disk (speed may be matters) to scan concentric “tracks”?
    For tapes, maybe by using also a floppy or HDD head to scan linearly a portion of tape, and then stitched every portions together to reconstitute the whold tape length?
    Having a kind of universal magnetic scanner with a software decoding would allow to recover content of any magnetic support without the need to rely on the original hardware aging electronics, missing controller, wrong device version…
    Maybe more difficult with analog signals, where signal is continuous.

    1. This already is done using non contact quantum magnitometer and further soctware signal procesing. You dont touch the tape but read it. As well with electron beam deflection.

      1. Kryoflux requires to use an original and working device, which could be somewhat difficult to provide and maintain. And it works only with floppy drives, not any other kind of magnetic format. And it samples only magnetic flux transition timings, so digital only, not analog.
        I’m thinking of something more universal, and not dependant of original/unobtainium devices.

    2. Except the signals on a floppy disk *are*, arguably, both analog and continuous. It’s the overall pattern and timing of the flux reversals on the disk that end up getting interpreted as a ‘0’ or a ‘1’ – the usual infographics that show data encoding as “magnetic domains pointing one way = 0, pointing the other way = 1” is incorrect.

      The fact that floppy disks are, at their core, still an analog medium, is the fundamental principle behind some of the copy protection that’s cropped up over the years: The most notable being “weak bits”, where a specific portion of a sector – or an entire sector or track – is recorded such that it doesn’t produce deterministic flux reversals after the signal from the head goes through the head amplifier. The resulting data ends up being different each time it’s read, and the program (usually a game) relies on this to determine that it’s running on a legitimate disk, since the average drive controller can’t physically write weak bits.

  2. Very cool. Similar to the IRENE project developed by the Lawrence Berkeley National Laboratory and the Library of Congress and used by the NEDCC. It takes close up photos of analog recordings (as opposed to CDs) and then is able to stitch them together and extract the audio out of the images.

  3. Back in ’96 Peter Gutmann, in a pretty astonishing USENIX paper, wrote about using Magnetic Force Microscopy to read disk drives from the physical media (he also wrote/writes about how to delete information, cryptography, and tons of other stuff.) As I recall he later opined that disks drives were getting too dense and the interference between data was making it too hard to read by similar methods, but there are always clever folx who figure out ways around simple laws of physics.

    If you can read the data from the media (directly or indirectly) you can reconstitute the files/filesystem with a number of methods used in the last 25+ years (from my own Lazurus to Binwalk and others.)

  4. I’ve wondered if you could read 78s using a scanner, especially as I broke one of my dad’s when I was a kid and he still has it. (Clean break in two, I accidentally leant on it).

  5. I’ve heard of this for reading vinal records. Skips and scratches can be removed from the data. An image of the record is used to reconstruct the audio. No needel neded to play the record.

Leave a Reply to Benjamin J SmithCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.