Minidisc FAQ: Disk Topics

  1. What are the dimensions of a MD?

    7 cm (2.75") x 6.75 cm (2 21/32") x 0.5 cm (3/16") The disc inside is 64 mm in diameter.

  2. What is the storage capacity of MDs, CDs, DATs, DCCs?

    MD: ~160MB for 74 min. in audio mode [140MB in Data Mode]
    CD-ROM: 650MB
    CD Audio: 740MB
    DAT: ~1GB
    DCC: 175MB [60 min. tape]

  3. What is the minimum allocation unit when recording on an MD?

    About 2 seconds for "SP" mode audio, 4 seconds for "SP" monaural, 4 seconds for "LP2" and 8 seconds for "LP4" mode audio. Audio information on Minidisc is organized in cluster units, which constitute the minimum unit for write or over-write of data to the disc. Each cluster contains 32 sectors of audio data, 3 "link sectors", and 1 sector of "sub data". Each sector is 2332 bytes, and 2 sectors are grouped together into a 4664 byte block holding 11 sound groups. Each sound group is 424 bytes and contains 11.6 msec of SP stereo data or 23.2 ms of SP mono data (LP2 and LP4 are double and quadruple the SP audio duration). A cluster is then 74624 bytes (32*2332), containing 176 sound groups (74624/424), for 2.0416 seconds of stereo sound (176*11.6) or 4.0832 seconds of monaural. See also the section on Prerecorded vs. recordable disc layout.

    The sound group (1/86th second) constitutes the finest resolution at which the audio stream can be edited. Older consumer grade MD decks have an edit positioning accuracy of approx. 60 msec (5 sound groups) but many modern machines allow edit positioning in units of a single sound group (11.6ms).

  4. What is the MD disc space allocation algorithm?

    According to Minidisc Rainbow Book recommendations (a ``remark'', not a mandatory requirement) the virgin space that runs to the end of the disc (if any) is always the first allocated, followed (in ascending disc address order) by the remaining free segments located throughout the disc. Free segments smaller than 6 clusters (one cluster is slightly more than 2 seconds of SP mode stereo) are not allocated until enough adjacent area is freed to make an open space of at least six whole clusters. (Thanks to David Tamkin for this investigation using a Sony MDS-JE520 MD deck.)

  5. Why is MD capacity always the same (74:59 exactly) regardless of how much redundancy there is in the recorded music?

    On a frame by frame (11.6 ms) basis, ATRAC is always trying to give as much resolution as it can to each spectral component that it deems important. However it is continually short of bits, and except for the case of zero input, or a very simple signal (e.g. a few pure tones), ATRAC is using all the bits per second it has available. If it had more bits per second available it could use them, and record the signal with better sound quality.

    In Sony's AES paper on ATRAC they talk about the bit allocation algorithm, and how it in general allocates more bits than are available, so they have a method for subtracting an offset from the allocations in order fit within their bit budget for a sound frame.

    In the zero or simple input case, there can be unused bits in some frames, but since there is an unvarying (i.e. fixed) relationship between 11.6 ms of audio and a 424 byte "sound group" on the disc, the system is not flexible enough to actually take up less disc space when it doesn't need it.

  6. What's the difference between the 60, 74 and 80 minute discs?

    The pregroove wobble is at a higher linear frequency (cycles/cm.) on the 74 minute blanks, causing the recorder to spin the disc slower (74 min: 1.2 m/s vs. 60 min: 1.4 m/s) and thereby record more information in the same linear space. The 80 minute discs are made by decreasing the track pitch (i.e. space between tracks) from the MD standard's 1.6um to 1.5um (micrometers), the minimum allowed by the MD specification; like the 74' blanks, they also run at 1.2 m/s. There is also information encoded in the read-only TOC near the inner circumference of the disc telling the player how long the disc is. Originally it was difficult to make the 74' blanks, but nowadays the manufacturing costs of 60' and 74' discs are the same (as well as, presumably, the 80's).

  7. Will MDs of the future contain un-compressed data?

    Sony produces high density Minidiscs and drives with a capacity of about 650MB for their MD Camcoder. Such capacity would allow a CD's worth of data to be stored uncompressed on an MD. However, new audio MDs made with such capacity would be incompatible with the existing hardware base (see below), and (assuming 16 bit linear recording), would not increase audio fidelity substantially. The main benefit of uncompressed recording would be the ability for professional users to make perfect MD duplicates. If the high density MD data drives become popular, it is possible that professional MD audio equipment will be made which uses MD data discs to store uncompressed audio.

  8. Will my MD player be able to play them?

    No, with their 4X increase in density the discs are physically incompatible with existing equipment.

  9. How does a pre-recorded MD differ from a recordable one?

    Physically, a pre-recorded MD is just like CD, using the same optical parameters, material, and production methods. Only the data contents are different (compressed ATRAC audio vs. 16 bit linear PCM audio). Unlike recordable MD, pre-recorded MDs do not have the magneto-optical coating layers or the lubricating layers. They are made with the same plastic-aluminum structure as CDs, so there is no way to record or erase anything on them. Prerecorded MD's are also read exactly like CD's (the player focuses a laser on pits and valleys within a transparent polycarbonate substrate backed by a coating of aluminum thus reflecting or dispersing the beam).

    Recordable MD's are similar, but a pre-groove replaces the pits and valleys and an MO coating replaces the aluminum one. When recording, a laser is focused from one side of the disc onto the pre-groove and heats a spot on the MO recording layer to its Curie point while a magnetic field from a head in contact with the other side of the disc aligns magnetic domains within (read: magnetizes) the heated spot on the MO layer (the N/S orientation corresponding to 0s and 1s in the data). During playback the MD machine focuses the laser on the pre-groove again, but at lower power, and the data is read back by measuring changes in polarization of light reflected from the previously magnetized regions (the Faraday effect).

    In order to playback both pre-recorded and recordable media, all MD units have a dual function optical assembly which can, depending upon the disc type, detect changes in either reflectivity or polarization.

    In terms of audio quality, pre-recorded MDs are in theory no different than recordable MDs, although, as always, audio quality depends upon which version of ATRAC the discs are encoded with. Some users have noticed certain pre-recorded MDs that sound worse than a homemade copy of the same CD with a modern MD recorder. It's likely that the ATRAC encoders used for prerecorded MDs are improved over time just as they are in the consumer units.

    Finally, recordable MDs have an extra hole in their case, which is sensed by a microswitch within the MD unit that connects to the REFLECT pin of the microprocessor. It tells the MD player's microprocessor whether signal reflectivity is high (for pre-recorded) or low (for recordable).

  10. Is there a difference in the disc layout of a recordable MD and pre-recorded MD?

    On recordable MDs, 32 out of 36 sectors in every cluster (the smallest recordable unit) are used for storing audio data. Of the remaining 4 sectors, 3 are needed to accommodate the start up and run out of the error correction code, as stated in the Minidisc IEEE paper: ``Because of the long interleaved ACIRC error correction code, three sectors must be used as "linking sectors." If the user changes or adds new data to the MD disc, two or three sectors for every start and end position of the new data need to be recorded.''. In other words, some space is wasted to accommodate re-recordability. On pre-recorded MDs, however, the data is fixed during mastering as one continuous stream, so these 3 extra sectors can be devoted to sub-data (but there is no documentation as to what this extra sub-data space could be used for). The Sony DADC MD handbook says that the maximum duration of a pre-recorded MD is 78:16.

  11. How many times can I record on a recordable MD?

    Sony claims a blank MD can handle up to 1 million recordings... Time will tell! In practice, the user Table of Contents (user TOC or UTOC) of an MD could become messed up by accident (defective MD recorder or physical damage) and render the MD unusable. The Sony portable MD recorder (MZ-1) allows to recover some unusable MD's in test mode (machine open and test jumper in place) via the TOC erase function: the TOC gets obliterated and that results in a "like-new" blank MD. Physical damage to the TOC cannot be repaired, but similar defects in other parts of the MD should be removed from the available space by creating short quiet tracks that include the problem and recording after labeling these as "bad".

  12. How long can the data/audio be stored on a recordable MD?

    Sony claims in their Magneto Optical FAQs that data may be stored with magneto optical technology for more than thirty years without loss or degradation. Once written to the disk, data are safe from the magnetic fields and heat found in normal environments. However, strong magnets placed directly against the MD can destroy data, as evidenced by informal experiments conducted by users. In contrast, a modern (ca. 1999) 'silver' pressed CD from a glass master is reckoned to last around 50 years or so from a chemical point of view. However a problem was found with the chemistry comprising early CD's (of around seven years and older) which caused degradation within a decade.

  13. What is the status of Sony's push to have MDs replace CD-ROMs?

    Sony wanted MD-Data to replace 3.5" floppy-disks (1.4 Mb, 2.8 Mb and 21 Mb floptical), not CD-ROM (a well implanted medium in the computer industry by now, and currently cheap). MD-Data's first incarnation was the MDH-10 SCSI drive, it provided 140 MB of data storage on an MD but was slow, expensive and incompatible with the Audio MD format (Data drives cannot read or write audio MD data except in "play" mode, which does not provide computer access to the data.). Sony subsequently released the MDM-111 for use in the IBM Thinkpad, but neither drive became popular and the MD-Data format has effectively been shelved, and appears only in certain niche products such as the MD still camera and MD document scanner.

  14. Can an MD Data drive be used to access the ATRAC data on an MD Audio disc?

    No, there is a read-only label track on each MD that distinguishes an MD Audio and MD Data disc, MD equipment looks at this label to determine what kind of disc it is accessing. An MD audio disc inserted in an MD Data drive shows up as having 1k in disc, 0k available. The ATRAC encoded music information is not accessible. An MD Data drive user has written some information about the MD Data drive and cross compatibility.

  15. Blank MDs are now available from several manufacturers. Do they all perform equally well? The only information available to me is the relative pricing so on what other basis can a reasoned choice of brand(s) be made?

    Since MDs store digital data, there will be no qualitative difference between the sound of various discs so long as the disc's raw bit errors fall within the bounds of what can be handled by the MD ACIRC error correction system.

    Presented here are the results of a small experiment I conducted to compare the raw block error rate of various brands.

    Blank Minidisc Raw Block Error Rates Compared

    Eric Woudenberg

    The service manual for the Sony MDS-503 Minidisc Deck tells how to put the unit into test mode, and from there how to display the BLER (raw block error rate [per second]) as it reads the disk. Using this feature I compared the block error rates of TDK, Idemitsu, Maxell, Panasonic, Sony and Keep blanks.

    In Andy Poggio's CD paper he says: "The CD specification allows for discs to have up to 220 raw errors per second. Every one of these errors is (almost always) perfectly corrected by the CIRC scheme for a net error rate of zero."

    I have reason to believe that 220 is the threshold for MD as well, since MD uses [modified] CD logic for encoding data on the disc, and since 220 is actually one of the error thresholds used for setting the MDS-503 focus bias.

    The un-surprising result: No discs had anywhere near an BLER of 220.

    The procedure is to use "continuous recording" in diagnostic mode to make a test disk, then look at the error rate while reading back from the disk.

    The table below shows the average and maximum BLERs for each brand. "In", "mid", and "out" refer to the three areas on the disc which are tested by default, they begin at cluster 40, 300, and 700 respectively.

    Minidisc Media Raw Block Error Rates

    Brand in mid out
    avg max avg max avg max
    TDK 4.97 20 5.08 15 4.16 19
    TDK (1) 4.05 22 5.00 15 3.66 18
    TDK (2) 5.96 20 2.47 6 3.84 18
    IDEMITSU (3) 2.11 15 1.15 5 1.22 5
    MAXELL 1.18 4 3.05 15 2.24 8
    PANASONIC 6.00 13 5.45 10 4.13 8
    SONY (4) 43.03 88 3.93 14 2.00 7
    KEEP 2.10 5 1.93 5 3.23 13
    Sony ES 1.60 5 1.00 3 1.59 5
    Canfield Audio 60/GL-16658 7.05 13 8.50 19 5.82 12
    Canfield Audio 74/GL-16658 11.76 31 10.13 25 9.36 20
    Fuji 60/ 5Z20M536 0.87 5 0.76 3 0.83 3
    (Original Data)
    1. Same TDK disc -- ejected and reinserted.
    2. Same TDK disc -- written once again with diagnostic.
    3. Disc contained a known bad spot (but not in test area).
    4. (Old) Gray Sony. Reported address errors.

    Some discs had been recorded on previously in normal audio recording ("non-continuous") mode. The BLER of these areas was typically 20-50, and always higher than the areas made with "continuous-recording" in diagnostic mode.

    The Sony disc was the only surprise, perhaps it had a weak/bad spot. In any case, it was still well within the presumed threshold for CIRC.

    I am not sure if comparative quality judgements can be made from these results since the tests were done on three short (approx. 10 second) intervals of each disc. I think the main conclusion is that all the discs are well within the threshold of "identical" from a post error correction point of view.

    The other consideration is longevity, and I quote two paragraphs from the excellent book by John Watkinson The Art of Digital Audio where he is discussing the magnetic layer of a Magneto-Optical disc:

    Magnetic layers with practical Curie temperatures are made from proprietary alloys of iron, cobalt, platinum, terbium, gadolinium and various other rare earths. These are all highly susceptible to corrosion in air and are also incompatible with the plastics used for moulded substrates. The magnetic layer must be protected by sandwiching it between layers of material which require to be impervious to corrosive ions but which must be optically transmissive. Thus only dielectrics such as silicon dioxide or alumnium nitride can be used.


    The master is developed and electroplated as normal in order to make stampers. The stampers make pre-grooved disks which are then coated by vacuum deposition with the MO layer, sandwiched between dielectric layers. The MO layer can be made less susceptible to corrosion if it is smooth and homogeneous. Layers which contain voids, asperities or residual gases from the coating process present a larger surface area for attack. The life of an MO disk is affected more by the manufacturing process than by the precise composition of the alloy.

    So, perhaps, more expensive discs might last longer (rather than have a lower initial error rate). I don't know the real situation though, maybe all the manufacturers have excellent quality control.

  16. I've heard that after many recordings and edits an MD can become fragmented, is this a problem?

    My comments are with reference to modern MD machines (ca. 1996), I am unfamiliar with the behavior of the older units. I am also speaking based upon my observation of simple experiments on an MDS-503, not on any intimate knowledge of MD technical specifications.

    There are two potential problems with fragmentation: interruption of music due to excessive seeking, and loss of free space.

    Concerning interruption of music, fragmentation alone will not cause a problem. A typical MD player can buffer 10 seconds of music and read the disc at 150 Kbytes/sec. (roughly the 1x CD data rate) which is over 4 times the MD audio rate. You will only have problems if it takes the player more time over any 10 second period to access and read segments of music than it does to play them.

    My simple tests with the '503 showed that it could not keep up with an arbitrarily long string of 2 second segments located at alternate ends of the disc (on a 74' MD), but that it could with 4 second segments. Assuming a full stroke seek takes about 2.3 seconds (measured by listening to the unit seek) and 3 seconds of music can be read in about 0.7 seconds, it would seem that the smallest segments a player could keep up with on a continual basis would be about 3 seconds long if they were located at worst case locations on the disc, and would cause the player to be constantly seeking and reading. You would need to have a string of segments this size or smaller, for at least 10 seconds duration, at opposite ends of the disc, to cause a problem.

    This simply cannot happen through fragmentation alone since all free list segments are at least 12 seconds long (see below). A worst case seeking condition that caused intermittent muting could still be created if 3 second or shorter segments from opposite ends of the disk were catenated together, but free list segments are allocated in sorted order (see FAQ section on allocation), so that unless you are trying to create such a pattern, you are rather unlikely to make one in the normal course of editing.

    The free space problem is caused by unused segments of disc space less than 12 seconds long that are not available for reuse. My simple tests showed that whenever the MDS-503 could coalesce space it did, so that lost space is only a problem when many small unused segments less than 12 seconds long are scattered throughout the disc, not adjacent to any existing free space. It is possible that in a normal editing operation of deleting dead space between tracks, up to 12 seconds per track could be lost, and on a disc with 25 tracks that would be 5 minutes lost to fragmentation. Though this does not seem troublesome, one real problem may occur in a sound effects application where all the tracks are tiny (< 12s), since if you were to delete every other track, nothing could be coalesced and there would be no change in the amount of free space.

    Though the degree of fragmentation depends upon your recording and editing patterns, I cannot see it causing problems in any but extreme circumstances. It should also be noted that all fragmentation is eliminated when an "Erase All" operation is performed. This restores the disc to a single TOC entry containing all the free space.

  17. MD Lens and Head cleaning discs are being sold, do these help?

    Two users have reported that their car units began skipping badly, but that after cleaning the laser lens by hand the problem disappeared (though both noted that the lens did not appear to have any dust or dirt on it before cleaning). Subsequently one user now uses a TDK lens cleaning disc when problems develop, which appears to be as effective as hand cleaning. No reports yet on the need for or effectiveness of the head cleaning discs. (Thanks to Jon Long ( and Shawn Lin (

    However, Arnaud Devilder (MPO MD disc sales manager) says: "I have to say that these products are only MARKETING. Avoid using them, we saw in our test labs that these products are very bad for your hardware."

  18. When a Sony MD deck (pre-MDS-JE520) is in record-pause mode, the disc is spinning and the magnetic head is in contact with it. How long can it safely be left in this state?

    [Note that this issue does not arise on portable recorders; in the interest of conserving battery life they do not spin the disc beyond what is necessary for reading and writing data. ] In their IEEE paper on the MD system, Sony writes:
    A very high durability of the [magnetic head] contact cycle can be achieved, e.g., more than a million passes.
    Taking the worse case scenario, an MD machine left in record-pause mode at the lead in (i.e. innermost) area of the disc [diameter: 32mm] and running at the highest linear velocity (1.4m/sec) would spin at [1.4/(.032*3.14159) * 60] = 836 rpm. At this speed, 1 million revolutions [passes] would take approx. 20 hours. So, record-pause for several hours is okay. A day or two is probably not.

    Here's a tip when recording a CD unattended: put the CD player in repeat-all, this will cause the MD player to record to the end and stop (rather than go into record-pause). You can easily delete the extra tracks later.

Copying Status: Public Domain. However, no liability is accepted for the correctness and safety any procedures or advice given here. This FAQ is distributed in the hope that it will be useful. Also, please include a notice that this file can be found at in any copies that you post.

Return to the MiniDisc Community Page.

As always, I would be much obliged for any updates and corrections from Sony Engineering. -Eric Woudenberg

Return to the Minidisc Community Page