Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!apple!sun-barr!texsun!pollux!killer!chasm
From: chasm@killer.DALLAS.TX.US (Charles Marslett)
Newsgroups: comp.sys.ibm.pc
Subject: Re: Question about diskette densities
Summary: 10 works sometimes!
Message-ID: <8465@killer.DALLAS.TX.US>
Date: 25 Jun 89 18:29:25 GMT
References: <1559@cbnewsh.ATT.COM> <586@megatek.UUCP> <8428@killer.DALLAS.TX.US> <7710@cbnews.ATT.COM>
Organization: The Unix(R) Connection, Dallas, Texas
Lines: 82

In article <7710@cbnews.ATT.COM>, cbema!las@cbnews.ATT.COM (cbema!las) writes:
> In article <8428@killer.DALLAS.TX.US> bobc@killer.DALLAS.TX.US (Bob Calbridge) writes:
> >In article <586@megatek.UUCP>, hollen@zeta.megatek.uucp (Dion Hollenbeck) writes:
> >+ From article <2210@bingvaxu.cc.binghamton.edu>, by vu0112@bingvaxu.cc.binghamton.edu (Cliff Joslyn):
> 
> >+ You didn't ask this, but it may help someone.  DSDD disks can be 
> >+ formatted for up to 800k.  This is an unusual configuration because
> >+ it uses 10 sectors per track which the UCSD p-System does reliably,
> >+ but I never figured out why MS-DOS only used a max of 9 per track.

I have had quite a bit of experience mailing diskettes, moving code from
one machine to another, etc, with 9 and 10 sector/track diskettes.  And
the 9 sector/track format works going from ATs to XTs, from Fujitsu drives
to Teac to Cannon to ...  On the other hand, 10 sector/track diskettes are
readable at the destination (usually), but the first time the dummy who
received the diskette tries to write anything to it - BOOM goes the FAT,
and nothing is accessable on the disk ever again!

(Appologies to anyone who has written a memo on a master disk -- I have done
it, too ;^)

The problem is that there is still some variation between drives (even
drives made the same year by the same company), and there is no room at
all for any speed deviation in the 10 sector/track format.

On the other hand, if you have reasonably well maintained drives and they
are no more than, say, two years old, you can probably use 10 sectors/track
reliably.  For backups, I reccommend you live with 9 or go to 15 (A 1.2 MB
diskette, that is to say -- they are not really very expensive, I pay $.50
each, and they are certainly worth that! [5 1/4s, 3 1/2s are still a bit more
expensive].

> In order to make room for 10 sectors per track on 5 1/4 inch floppy
> disks, you must reduce the gap between sectors to a value less than
> the "standard" originally defined by IBM (let's see... I think they
> called it 3740, yeah that sounds about right) for floppy disks.  Of
> course, that standard was very conservative, having been tailored to
> the capabilities of their very early tech 8 inch floppies (called -
> surprise, surprise - the 3740).  

The original gap was 42 bytes (out of about 330) for 8" media, for 5 1/4"
media it was increased to 48 bytes (the drive power supplies and moters
were not so reliable), and to add a tenth sector it has to be set to
around 15 or so.  Since the floppy data seperator requires 10 bytes or
so, this leaves very little room for speed variation (5 bytes out of 300
is less than 2%).

> Advances in the state of the art had long since reduced the need for 
> such conservative specs when IBM decided to go from 8 to 9 sectors 
> per track on the PC, but apparently the old spec was too much of a 
> barrier to cross.  It is, after all, an IBM standard.  It took a non-
> IBM maverick company to go to 10 sectors per track, but the P-System 
> was never really in the PC mainstream and it didn't have much of an
> affect on the DOS mainstream.

IBM did not change anything when they went from 8 to 9 sectors/track,
since the difference was a performance and simplicity loss on one side
in exchange for a capacity gain on the other (old drives stepped very
slowly, and with essentially an unused sector at the end of each track
the drive had time to step to the next track before its first sector came
under the read/write head.  Today we use skewed formatting if we need that
kind of tweaking, and we still get the capacity.  The format routine is
just a bit more complex.

It had no effect on the DOS mainstream, mostly because it would never have
been accepted by software vendors, so anyone who used it would have to
support two formats (read, more work, no more pay, never get done ;^).
No major vendor thought it worth the effort.

> regards, Larry
> -- 
> Signed: Larry A. Shurr (cbema!las@att.ATT.COM or att!cbema!las)
> Clever signature, Wonderful wit, Outdo the others, Be a big hit! - Burma Shave
> (With apologies to the real thing.  The above represents my views only.)
> (Please note my mailing address.  Mail sent directly to cbnews doesn't make it.)


===========================================================================
Charles Marslett
STB Systems, Inc.  <== Apply all standard disclaimers
Wordmark Systems   <== No disclaimers required -- that's just me
chasm@killer.dallas.tx.us