Floppy disks again

From: Bill Yakowenko <yakowenk_at_cs.unc.edu>
Date: Thu Oct 14 02:35:31 1999

I wrote:
> How is it possible that DD media could be of such
> poor quality that it can't (reliably?) do 96 TPI,
> while still being just fine at 48 TPI?

To which Allison replied:
] ???? Nonsense question, no context. I have no difficulty
] nor is there any reason to expect difficulty with 96tpi DD
] ops. I have expereinced using formerly formatted media of
] 96 or 48TPI that REQUIRED bulk erasure to be usable. I believe
] that intertrack noise due to differnt track widths are why.
] I regurally use PC360k, Vt180, RX50 and Visual1050 media and they
] are 48 TS, 48 SS, 96ss, 96ss/TS all running at DD data rates.
...much more, deleted...


Oh yeah, you can't read my mind. Guess I better include some
context.

I'm thinking of the 5.25" DS/DD floppy disks commonly used
in PC's and formatted to hold 360K at 48TPI. I was recently
discussing with someone whether or not they could be used as
RX50's, which as you know have 96 tracks per inch, with each
track having the same number of bits as in the PC 360K DS/DD
format. Both are supposed to be 300 Oerstedts. But I think
I remember hearing complaints that the not-so-high-quality
DS/DD 360K floppy disks would not work okay with the higher
number of tracks per inch. And I am sure I read something
about the importance of these diskettes being "well honed"
to be used as RX50's, with the assertion that not all 360K
DS/DD disks are. (But "well honed" was left undefined, and
I'd love to know what exactly was meant, as well as what
physical property of those not-so-high-quality diskettes
was causing them to fail.)

Now I'm not talking about writing more on any one track. It
might still be just exactly the same as the PC DS/DD format.
And coercivity doesn't come into the picture, because we are
talking about floppy drives that are made to work with 300
Oerstedt media, which those DS/DD disks are.

I'm also not thinking of taking a diskette written by one drive
and trying to read it in another. That can cause problems
because 48TPI tracks might be twice as wide as those written
in a 96TPI drive. Simple enough; no confusion.

The point I'm after is about the limitation on how many
tracks that disk can hold - does that limitation come from
something about the diskette itself, or is it a property
of the disk drive in which it is used? My thinking is that
the media can handle >3000 magnetic transitions (bits) per
inch, because it does that along each track. So that can't
be the limiting factor that prevents you from fitting 96
tracks into one inch. I'm not concerned with the drive in
which it is written; I only care about what the media could
hold if I had a precise enough disk drive.

Assuming that the magnetic granules in the oxide are not
systematically shaped in any odd way, (not generally longer
than they are wide, or wider than long), then the media
itself should in principle be able to handle 3000 tracks
per inch - just as many transitions moving outward from
center as moving along a track. If the diskette itself
was so crappy that it couldn't handle 96 magnetic
transitions radially, then it should have no hope of being
able to handle 3000/inch when writing a track, so it would
also have no hope of being usable in the PC 360K 48TPI format.
Conversely, if it could handle the 3000 transitions per inch
of the PC 360K DS/DD format, then the media itself is already
*much* better than necessary to handle 96TPI.

So the limitation is in the drive, and not the diskette;
and any old never-written (or properly bulk-erased)
"PC 360K" diskette should be perfectly happy as an RX50.

It just occurred to me that the recorded track must be
more or less like a UPC bar code - the width of the track
is very much greater than the length of a bit in the track;
maybe more than 200 times greater. So say the track was a
1 meter wide sidewalk, the bits would each be 1m wide, but
only 5mm long. So even a tiny twist in the read head would
gum it up real good - it should be much more sensitive to
this than to alignment. Move 1cm to the left, and you are
still 99% on the sidewalk - no problem. But twist just one
degree, and when your left side is at one bit, your right
side is three bits ahead or behind - big trouble. Well,
maybe the forces acting on the head are less likely to
twist it than to shift it.

        Bill.
Received on Thu Oct 14 1999 - 02:35:31 BST