Image Map Image Map
Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 21

Thread: MFM and Double Density

  1. #11
    Join Date
    Feb 2006
    Location
    in the basement
    Posts
    764

    Default

    Thank you all for your brilliant comments:


    ziloo

  2. #12

    Default

    Quote Originally Posted by ziloo View Post
    Alrighty then....let me rephrase my statement; comparing two modes of
    modulation, namely FM and MFM, we can say that on the same media
    the shortest safe distance between two pulses (flux-transitions) is that
    of an FM........Yes....NO...?

    ziloo
    Reread my post. They get more data by removing the unneeded clock pulses on MFM. The data bits are sufficient to keep the reading PLL locked on phase. There is no increases in the density of transition on the disk. MFM just doesn't waste space for clock pulses that are not needed.
    Clocks are only added when there are empty strips with no data transitions.
    FM puts a clock pulse in for every data, even when the data doesn't change.
    Dwight

  3. #13
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,233
    Blog Entries
    20

    Default

    I should add this:

    There are more ways than simple FM or MFM to getting more bang for the buck, while still remaining within the limits of bandwidth and the need to preserve synchronization. One such way is "Group Code Recording", or GCR. Basically, the idea is to encode groups of bits into a larger group, selecting optimal bit sequences. This "more is less" approach works quite well. It was used by Apple, Victor, Micropolis, Commodore and others. Consider:

    (1) You can have any number of reversals in a row, but
    (2) You can't have two periods or more of non-reversal in a row (you'll lose synchronization).

    So, create the following table of bit sequences:

    10101010
    10101011
    10101110
    10101111
    ...
    11111011
    11111101

    For a total of 32 combinations or five bits' worth of data. Note that, unlike FM or MFM, there are no separate clock bits--the data is self-clocking.

    You can see this sort of thing on inexpensive word processors, where the logic for MFM reading and writing is a cost concern, but FM doesn't give sufficient density.

    Similarly, the Apple II used what amounts to a latch, shift register and ROM to implement a floppy controller using group-code (see the "Inside the Apple II" book for details; there were two schemes used). Commodore used approximately the scheme described above.

    There are a lot of variations on the theme. Some can tolerate two or more periods of non-reversals; this enables more efficient coding at the expense of slightly less reliability.

    GCR on floppies largely fell by the wayside after standard LSI floppy controllers became more affordable. What was once a major cost item that required extensive support circuitry went to occupying a corner of a SuperIO chip. Moore's law again.

    Group code on tape drives has been around since 1973. The GCR scheme used on 6250 cpi tapes is quite clever. Later drives take the idea further with better bit densities.
    Last edited by Chuck(G); December 6th, 2017 at 06:46 AM.

  4. #14
    Join Date
    Feb 2006
    Location
    in the basement
    Posts
    764

    Default

    Sorry folks, my last question was very ill posed and caused
    so much confusion.
    Let's just delete that for now.


    ziloo

  5. #15
    Join Date
    Feb 2006
    Location
    in the basement
    Posts
    764

    Default

    As I understand, there were times when data to be copied on
    the floppy disks were "inverted":

    (the bottom image is ...em...like...ummmm .... kinda the inverted image of
    the top one ..........well that was best emoji I could find... )

    that is 11010010 would be recorded as 00101101. What was the reason for that?


    ziloo

  6. #16
    Join Date
    May 2009
    Location
    Connecticut
    Posts
    3,446
    Blog Entries
    1

    Default

    Look up the difference between big-endian and little-endian. Having the order of bytes stored on disk match the order of bytes in memory makes everything just a little faster.

  7. #17
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,233
    Blog Entries
    20

    Default

    Floppy disks have traditionally used the opposite in bit ordering within a byte from, say, USART ordering. There are, as with anything having to do with floppies, exceptions, so it's not a given. It doesn't really matter; just a convention.

  8. #18
    Join Date
    Feb 2015
    Location
    Cleveland, OH, USA
    Posts
    501

    Default

    Don't mean to confuse the discussion but Chuck's comment about group-code schemes reminded me that Commodore also used zone bit recording (variable sector counts) as another scheme for packing more data onto a disk. The idea was that the tracks are larger at the outer edge of the media so they created more sectors there. Put another way, the physical limit of density on the disk is found at the inner rings (tracks) near the hub.

    See: https://en.wikipedia.org/wiki/Zone_bit_recording

    -CH-

  9. #19
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    25,233
    Blog Entries
    20

    Default

    That zoned recording technique was also used in the Victor/Sirius and the Twiggy drives used a variation of it, varying from 15 to 22 sectors per track. (both of which also used GCR, a separate technique not related to zoned recording).

    The scheme, of course, is much older than PCs; it was described in literature for the Bryant 4000 hard disk, manufactured by everyone's favorite milk carton maker...

  10. #20
    Join Date
    Feb 2015
    Location
    Cleveland, OH, USA
    Posts
    501

    Default

    I thought they made sponges...

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •