IBM 3740 Track Format is the soft-sectored FM format used by Shugart SA-800 drives and descendants. An overview of the format can be found in this document from Shugart, starting on PDF page 11.
In order for a controller to figure out where the floppy is relative to a full rotation of the disc, a controller will look for specific violations of FM encoding rules known as Address Marks (specifically clocking bits will be removed which should otherwise always be there).
This track format is used by PC floppy controllers and descendants from the original Shugart interface. As I understand it, a controller is able to sync to the bits being read by detecting specific violations of MFM encoding that precede the Address Marks known as A1/C2 sync. The Address Marks themselves follow MFM encoding correctly. This is described in detail in Western Digital's FD1791 datasheet.
However, I recently read another document from Shugart, and it provides an alternate form of synchronization, starting on PDF Page 9. Shugart's format suggests there are no A1/C2 sync marks, and Address Marks again are the coding violations, just like in the MFM format.
My own tests reading from a 765-formatted disc are consistent with Western Digital's results. Since parts of the track WD and Shugart use (as well as their corresponding bit patterns) for synchronization are different, I can't reconcile the results I'm seeing in my tests with the Shugart document I'm reading. Did Shugart use it's own specific MFM format, and WD used a different one? Can anyone else confirm there were multiple variants to System 34? Which is the "correct" encoding?
In order for a controller to figure out where the floppy is relative to a full rotation of the disc, a controller will look for specific violations of FM encoding rules known as Address Marks (specifically clocking bits will be removed which should otherwise always be there).
This track format is used by PC floppy controllers and descendants from the original Shugart interface. As I understand it, a controller is able to sync to the bits being read by detecting specific violations of MFM encoding that precede the Address Marks known as A1/C2 sync. The Address Marks themselves follow MFM encoding correctly. This is described in detail in Western Digital's FD1791 datasheet.
However, I recently read another document from Shugart, and it provides an alternate form of synchronization, starting on PDF Page 9. Shugart's format suggests there are no A1/C2 sync marks, and Address Marks again are the coding violations, just like in the MFM format.
My own tests reading from a 765-formatted disc are consistent with Western Digital's results. Since parts of the track WD and Shugart use (as well as their corresponding bit patterns) for synchronization are different, I can't reconcile the results I'm seeing in my tests with the Shugart document I'm reading. Did Shugart use it's own specific MFM format, and WD used a different one? Can anyone else confirm there were multiple variants to System 34? Which is the "correct" encoding?
Last edited: