Image Map Image Map
Results 1 to 8 of 8

Thread: MFM / RLL encoding on old hard drives

  1. #1

    Default MFM / RLL encoding on old hard drives

    Hi,

    I have an old 20MB MFM hard drive (https://stason.org/TULARC/pc/hard-dr...MFM-ST412.html) and I'm currently experimenting with it on different controllers cards. Most of these cards are MFM controller cards.
    I also have one RLL controller card and much to my surprise (before reading up on the subject) I noticed that my 20MB hard all of a sudden became a 30MB hard drive (makes sense now after doing some reading)

    Now if I understand correctly, on a hardware level there isn't really much of a difference between a drive that uses MFM encoding or RLL encoding (it is the way the BIOS lays out the tracks during the low level format).
    According to https://redhill.net.au/d/10.php : "RLL did, however, require a very accurate drive mechanism"
    It also states that "Not many early drives, however, were capable of taking an RLL format reliably.". How does that manifest itself on non-RLL capable drives (like mine probably).

    • Will that result in lots of bad sectors after formatting ?
    • Will that result in data loss along the way (read / write errors) ?
    • Can I recover from those errors by switching back to MFM (re-formatting the drive with an MFM controller), or can this cause permanent damage to the drive
    • Is there an exhaustive list somewhere of drives that are capable of "safely" doing RLL encoding ?


    Thx.

  2. #2
    Join Date
    Mar 2011
    Location
    Atlanta, GA, USA
    Posts
    1,421

    Default

    Quote Originally Posted by RetroSpector78 View Post
    ...on a hardware level there isn't really much of a difference between a drive that uses MFM encoding or RLL encoding...
    • Will that result in lots of bad sectors after formatting ?
    • Will that result in data loss along the way (read / write errors) ?
    Yes and no. There is a difference in bit level encoding but the main reason you get more capacity is the RLL controllers bit-clock is 50% faster at 7.5 MHz vs 5.0 MHz for MFM. This results in a higher bit density on the magnetic medium. Lots of factors combine to develop CRC errors over time. One big factor is how closely the flux transitions live to their neighbors. Many drive manufacturers bin'd the grade of oxide coating curing QA to decide of those platter batches went in MFM or RLL part numbers. Writing a denser RLL encoding on a MFM graded platter isn't the dominant factor for errors, but depending on the age, manufacturing processes, lifetime usage, batch quality, etc, it may be a significant number of rain drops contributing to the flood.

    Quote Originally Posted by RetroSpector78 View Post
    • Can I recover from those errors by switching back to MFM (re-formatting the drive with an MFM controller), or can this cause permanent damage to the drive
    • Is there an exhaustive list somewhere of drives that are capable of "safely" doing RLL encoding ?
    Yes, you can always switch back with a re-low-level format. It won't harm the drive. Any list would have to contain manufacturing dates and batch numbers to be valid as there was a lot of variance in factors that contributed to weak flux intensity besides just density.
    "Good engineers keep thick authoritative books on their shelf. Not for their own reference, but to throw at people who ask stupid questions; hoping a small fragment of knowledge will osmotically transfer with each cranial impact." - Me

  3. #3
    Join Date
    Jan 2013
    Location
    Marietta, GA
    Posts
    3,150

    Default

    What will usually happen when you format an MFM drive with RLL is that it will look like it formated OK or with few errors, and then as you fill up the inner tracks with real data, hundreds of read errors will pop up. (This is not physical damage - reformatting with an MFM card will set things back to normal)

    However some MFM drives will format reliably to RLL depending on how they were manufactured. The easiest way to tell is to run Spinrite immediately after formatting the drive. If its thorough pattern test succeeds with no or only a few bad sectors, then it should keep working. Otherwise you will see piles of "bad" sectors starting about half way through the testing.

  4. #4
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    31,483
    Blog Entries
    20

    Default

    While RLL results in a higher recovered data rate, the spectrum on disk has the same high-frequency limit that ordinary MFM (and for that matter, FM) does. If you've got a good spectrum analyzer, you can verify this for yourself.

    (2,7) RLL (and for that matter, any run-length-limited code) gains information density by being more critical of the exact timing of domain reversals.

    Consider ordinary FM, for example--you've got an explicit clock bit in each cell, so exact timing isn't all that critical--you always (well almost, but that's a different matter) have a clock bit and then either you have a data bit transition (a '1') or you don't (a "0"). MFM doubles the information density without putting demands on the high-frequency limit by deriving the clock from the position (or lack of it) of a transition. It's less robust than FM in terms of speed variations and media characteristics.

    RLL similarly relies on more accurate (than MFM or FM) transition timing, which can be affected by media coating as well as spindle motor speed control. That's why most drives certified for RLL performance use plated media rather than an emulsion (an emulsion is thicker with large particles, so transitions aren't as accurate). That's not to say that hard drives whose platters are coated with rusty goo, rather than plating can't do (2,7) RLL (or even (3,11) RLL), but they're less suited to the application.

  5. #5

    Default

    Quote Originally Posted by SomeGuy View Post
    What will usually happen when you format an MFM drive with RLL is that it will look like it formated OK or with few errors, and then as you fill up the inner tracks with real data, hundreds of read errors will pop up. (This is not physical damage - reformatting with an MFM card will set things back to normal)

    However some MFM drives will format reliably to RLL depending on how they were manufactured. The easiest way to tell is to run Spinrite immediately after formatting the drive. If its thorough pattern test succeeds with no or only a few bad sectors, then it should keep working. Otherwise you will see piles of "bad" sectors starting about half way through the testing.
    Thx a lot ... will try the Spinrite pattern test ... Spinrite has been on my todo-list for a long time ... Can you recommend other tools to verify hard drive capabilities, or benchmarking tools for these mfm drives ?

  6. #6
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    31,483
    Blog Entries
    20

    Default

    If you're serious, there were hard drive "exercisers" or "testers" that did away with the controller entirely. Since an ST506 drive has no smarts on the data lines, it's a pretty simple matter of feeding a differential pulse train to the write electronics and then reading it back and comparing the recovered pulse train with what was written.

    But I suspect that level of detail isn't what you're looking for.

  7. #7

    Default

    SpinRite II version 2.0 is what he wants.
    PM me if you're looking for 3" or 5" floppy disks. EMail For everything else, Take Another Step

  8. #8
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    31,483
    Blog Entries
    20

    Default

    Yabut it's as much a test of the controller as it is of the drive. If you want an objective test of the drive, then it's a special hardware rig.

Tags for this Thread

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
  •