Floppy format problem.

Floppy format problem.

Post by David . » Mon, 24 Apr 2000 04:00:00




> Has anyone successfully formatted a 5.25" double-density 360K
> floppy diskette on a high-density 1.2 Meg floppy drive under
> Linux, specifically RedHat 6.0?  Or any version of Linux?

> I use the command:
>   fdformat /dev/fd0d360

> and get the following:

>   Double-sided, 40 tracks, 9 sec/track. Total capacity 360 kB.
>   Formatting ... done
>   Verifying ... end_request: I/O error, dev 02:04 (floppy), sector 0
>   Read: : Input/output error
>   Problem reading cylinder 0, expected 9216, read -1

> 2
>   end_request: I/O error, dev 02:04 (floppy), sector 4
>   end_request: I/O error, dev 02:04 (floppy), sector 6
>   end_request: I/O error, dev 02:04 (floppy), sector 9
>   <etc>

>   I don't have this problem when formatting a high-density 1.2 Meg
> floppy under Linux, and Win 98 will format either 360K or 1.2 Meg
> without a problem.  I've also tried a different floppy drive
> with the same results.  (Both are Teac drives; one is about 10 years
> old and the other is new.)

> Any suggestions as to what I might be doing wrong?  Or is this a
> bug?

> Regards,


You might try   fdformat /dev/fd0H360

--
Registered with the Linux Counter.  http://counter.li.org
ID # 123538

 
 
 

Floppy format problem.

Post by Charles Sulliva » Tue, 25 Apr 2000 04:00:00


Has anyone successfully formatted a 5.25" double-density 360K
floppy diskette on a high-density 1.2 Meg floppy drive under
Linux, specifically RedHat 6.0?  Or any version of Linux?

I use the command:
  fdformat /dev/fd0d360

and get the following:

  Double-sided, 40 tracks, 9 sec/track. Total capacity 360 kB.
  Formatting ... done
  Verifying ... end_request: I/O error, dev 02:04 (floppy), sector 0
  Read: : Input/output error
  Problem reading cylinder 0, expected 9216, read -1

2
  end_request: I/O error, dev 02:04 (floppy), sector 4
  end_request: I/O error, dev 02:04 (floppy), sector 6
  end_request: I/O error, dev 02:04 (floppy), sector 9
  <etc>

  I don't have this problem when formatting a high-density 1.2 Meg
floppy under Linux, and Win 98 will format either 360K or 1.2 Meg
without a problem.  I've also tried a different floppy drive
with the same results.  (Both are Teac drives; one is about 10 years
old and the other is new.)

Any suggestions as to what I might be doing wrong?  Or is this a
bug?

Regards,


 
 
 

Floppy format problem.

Post by Gene Hesket » Tue, 25 Apr 2000 04:00:00


Unrot13 this;

Gene Heskett sends Greetings to Charles Sullivan;

 CS> Has anyone successfully formatted a 5.25" double-density 360K
 CS> floppy diskette on a high-density 1.2 Meg floppy drive under
 CS> Linux, specifically RedHat 6.0?  Or any version of Linux?

 CS> I use the command:
 CS>   fdformat /dev/fd0d360

 CS> and get the following:

 CS>   Double-sided, 40 tracks, 9 sec/track. Total capacity 360 kB.
 CS>   Formatting ... done
 CS>   Verifying ... end_request: I/O error, dev 02:04 (floppy), sector 0
 CS>   Read: : Input/output error
 CS>   Problem reading cylinder 0, expected 9216, read -1

 CS>  sector
 CS> 2
 CS>   end_request: I/O error, dev 02:04 (floppy), sector 4
 CS>   end_request: I/O error, dev 02:04 (floppy), sector 6
 CS>   end_request: I/O error, dev 02:04 (floppy), sector 9
 CS>   <etc>

 CS>   I don't have this problem when formatting a high-density 1.2 Meg
 CS> floppy under Linux, and Win 98 will format either 360K or 1.2 Meg
 CS> without a problem.  I've also tried a different floppy drive
 CS> with the same results.  (Both are Teac drives; one is about 10 years
 CS> old and the other is new.)

 CS> Any suggestions as to what I might be doing wrong?  Or is this a
 CS> bug?

1. It may be a bug.

2. It may be that the linux writers looked at that bad situation and
decided not to cover it.  I sure wouldn't, its a disaster looking for a
place to happen.

The reasons are pretty basic.

a: the tracks laid by that DSQD drive are only about 40% of the width of a
true DSDD 360k drive, meaning there is gonna be huge amounts of noise
from the old data on either side of the new "formatted" track.

b: In addition to double-stepping the heads to make it into a 40 track
drive, a 48 tpi if you will, the spindle speed must be slowed down from
360 to 300 rpm, and the data pump must be switched down from 500
kilobaud to 250 kilobaud.

c: to go along with the spindle speed, data rate changes, the recording
currents applied to the heads should also be reduced IF its a DSDD disk,
something the drive has no sensors to detect, otherwise the heads would
be driven so hard that the recording is actually partially erased by the
excess currant.

Overall, trying to make a 360k diskette in a 1.2 megger will have to
deal with the fact that its probably can only be read in that same
drive, no real 360k drive, with its wider heads, would ever give you an
error free read.

Did I mention its a bad idea?  The only snowballs chance of making it
work is if you started with a bulk erased diskette, one erased by one of
those $30 tape erasers they sell at radio shack.  You could probably put
that disk in a real 360k drive and read it, but if you ever wrote to it,
and then tried to re-write that same area again in the QD drive, at
least that file would be trashed forever.

Cheers, Gene
--



                        email gene underscore heskett at iolinc dot net
This messages reply content, but not any previously quoted material, is
? 2000 by Gene Heskett, all rights reserved.
--

 
 
 

Floppy format problem.

Post by David » Tue, 25 Apr 2000 04:00:00



> Has anyone successfully formatted a 5.25" double-density 360K floppy
> diskette on a high-density 1.2 Meg floppy drive under Linux,
> specifically RedHat 6.0?  Or any version of Linux?

> I use the command:
>   fdformat /dev/fd0d360

use /dev/fd0h360

        d360 is for 360K drives
        h360 is for 360K disks in 1.2M drives

man fd(4) for more information.

-- David

 
 
 

Floppy format problem.

Post by David » Tue, 25 Apr 2000 04:00:00



> You might try   fdformat /dev/fd0H360

See the man page for fd(4).

H360 is for formatting a 360K disk in a 1.44M 3.5" drive.

You want h360 for a 360K disk in a 1.2M 5.25" drive.

-- David

 
 
 

Floppy format problem.

Post by Charles Sulliva » Tue, 25 Apr 2000 04:00:00




> > Has anyone successfully formatted a 5.25" double-density 360K floppy
> > diskette on a high-density 1.2 Meg floppy drive under Linux,
> > specifically RedHat 6.0?  Or any version of Linux?

> > I use the command:
> >   fdformat /dev/fd0d360

> use /dev/fd0h360

>         d360 is for 360K drives
>         h360 is for 360K disks in 1.2M drives

That's it David.  fd0h360 does the job.  Thanks.

Regards,
Charles Sullivan

Quote:> man fd(4) for more information.

> -- David

 
 
 

Floppy format problem.

Post by Charles Sulliva » Tue, 25 Apr 2000 04:00:00



> Unrot13 this;

> Gene Heskett sends Greetings to Charles Sullivan;

>  CS> Has anyone successfully formatted a 5.25" double-density 360K
>  CS> floppy diskette on a high-density 1.2 Meg floppy drive under
>  CS> Linux, specifically RedHat 6.0?  Or any version of Linux?

>  CS> I use the command:
>  CS>   fdformat /dev/fd0d360

>  CS> and get the following:

>  CS>   Double-sided, 40 tracks, 9 sec/track. Total capacity 360 kB.
>  CS>   Formatting ... done
>  CS>   Verifying ... end_request: I/O error, dev 02:04 (floppy), sector 0
>  CS>   Read: : Input/output error
>  CS>   Problem reading cylinder 0, expected 9216, read -1

>  CS>  sector
>  CS> 2
>  CS>   end_request: I/O error, dev 02:04 (floppy), sector 4
>  CS>   end_request: I/O error, dev 02:04 (floppy), sector 6
>  CS>   end_request: I/O error, dev 02:04 (floppy), sector 9
>  CS>   <etc>

>  CS>   I don't have this problem when formatting a high-density 1.2 Meg
>  CS> floppy under Linux, and Win 98 will format either 360K or 1.2 Meg
>  CS> without a problem.  I've also tried a different floppy drive
>  CS> with the same results.  (Both are Teac drives; one is about 10 years
>  CS> old and the other is new.)

>  CS> Any suggestions as to what I might be doing wrong?  Or is this a
>  CS> bug?

> 1. It may be a bug.

> 2. It may be that the linux writers looked at that bad situation and
> decided not to cover it.  I sure wouldn't, its a disaster looking for a
> place to happen.

> The reasons are pretty basic.

> a: the tracks laid by that DSQD drive are only about 40% of the width of a
> true DSDD 360k drive, meaning there is gonna be huge amounts of noise
> from the old data on either side of the new "formatted" track.

> b: In addition to double-stepping the heads to make it into a 40 track
> drive, a 48 tpi if you will, the spindle speed must be slowed down from
> 360 to 300 rpm, and the data pump must be switched down from 500
> kilobaud to 250 kilobaud.

> c: to go along with the spindle speed, data rate changes, the recording
> currents applied to the heads should also be reduced IF its a DSDD disk,
> something the drive has no sensors to detect, otherwise the heads would
> be driven so hard that the recording is actually partially erased by the
> excess currant.

> Overall, trying to make a 360k diskette in a 1.2 megger will have to
> deal with the fact that its probably can only be read in that same
> drive, no real 360k drive, with its wider heads, would ever give you an
> error free read.

> Did I mention its a bad idea?  The only snowballs chance of making it
> work is if you started with a bulk erased diskette, one erased by one of
> those $30 tape erasers they sell at radio shack.  You could probably put
> that disk in a real 360k drive and read it, but if you ever wrote to it,
> and then tried to re-write that same area again in the QD drive, at
> least that file would be trashed forever.

> Cheers, Gene

Gene,
It turns out that using fd0h360 (as suggested by David C) instead
of fd0d360 works, and if I then use:
  mkfs -t msdos /dev/fd0h360
I get a diskette which is compatible with DOS/Win 98.

What you've dscribed out would have merit were it not for the fact
that the usual 1.2 Meg drives are designed to at least read 360K
double density floppies, and I was able to successfully format one
under DOS/Win 98 (dual boot on the same box).  The drives change
speed and data rate depending on the density.

What is undoubtedly true is that this diskette will be unreadable in a
standard 360K double density drive with its wider heads.  I think
my 10 year-old drive, a Teac FD55GFR-142, had some sort of dual heads
which allowed writing either wide or narrow tracks, but I can't
find the specs for it on Teac's website.  (I'm pretty sure I used
360K floppies to move files back and forth between one of the antiques
I used to play with and the PC with the Teac -142.)  The specs for the
new one, FD55GFR-7193, don't mention this.  I trashed my last antique
with a double density 5.25" drive about 5 years ago so can't check
it out.

I appreciate your taking the time to respond.

Regards,
Charles Sullivan

 
 
 

Floppy format problem.

Post by David » Wed, 26 Apr 2000 04:00:00


Charles Sullivan writes:

>> 2. It may be that the linux writers looked at that bad situation and
>> decided not to cover it.  I sure wouldn't, its a disaster looking for
>> a place to happen.

>> The reasons are pretty basic.

>> a: the tracks laid by that DSQD drive are only about 40% of the width
>> of a true DSDD 360k drive, meaning there is gonna be huge amounts of
>> noise from the old data on either side of the new "formatted" track.

>> b: In addition to double-stepping the heads to make it into a 40
>> track drive, a 48 tpi if you will, the spindle speed must be slowed
>> down from 360 to 300 rpm, and the data pump must be switched down
>> from 500 kilobaud to 250 kilobaud.

>> c: to go along with the spindle speed, data rate changes, the
>> recording currents applied to the heads should also be reduced IF its
>> a DSDD disk, something the drive has no sensors to detect, otherwise
>> the heads would be driven so hard that the recording is actually
>> partially erased by the excess currant.

>> Overall, trying to make a 360k diskette in a 1.2 megger will have to
>> deal with the fact that its probably can only be read in that same
>> drive, no real 360k drive, with its wider heads, would ever give you
>> an error free read.

>> Did I mention its a bad idea?  The only snowballs chance of making it
>> work is if you started with a bulk erased diskette, one erased by one
>> of those $30 tape erasers they sell at radio shack.  You could
>> probably put that disk in a real 360k drive and read it, but if you
>> ever wrote to it, and then tried to re-write that same area again in
>> the QD drive, at least that file would be trashed forever.

> It turns out that using fd0h360 (as suggested by David C) instead
> of fd0d360 works, and if I then use:
>   mkfs -t msdos /dev/fd0h360
> I get a diskette which is compatible with DOS/Win 98.

> What you've dscribed out would have merit were it not for the fact
> that the usual 1.2 Meg drives are designed to at least read 360K
> double density floppies, and I was able to successfully format one
> under DOS/Win 98 (dual boot on the same box).  The drives change speed
> and data rate depending on the density.

> What is undoubtedly true is that this diskette will be unreadable in a
> standard 360K double density drive with its wider heads.  I think my
> 10 year-old drive, a Teac FD55GFR-142, had some sort of dual heads
> which allowed writing either wide or narrow tracks, but I can't find
> the specs for it on Teac's website.  (I'm pretty sure I used 360K
> floppies to move files back and forth between one of the antiques I
> used to play with and the PC with the Teac -142.)  The specs for the
> new one, FD55GFR-7193, don't mention this.  I trashed my last antique
> with a double density 5.25" drive about 5 years ago so can't check it
> out.

If I remember correctly from my DOS days....

A 360K disk in a 1.2M drive is written by double-writing the tracks.
That is, the head writes two adjacent identical tracks for each track
written by the software.  This way, the result is wide enough for a 360K
drive to read.

The effect isn't perfect, however, and some 360K drives do have problems
reading disks that are formatted in 1.2M drives, but it works most of
the time.  (I never had a problem with 360K drives reading disks
formatted in my Teac 1.2M drive, but I have had problems with 360K disks
formatted in an old IBM PC/AT.)

I'm not sure if all this is done by the drive and controller (after
receiving some kind of mode-setting command) or by software, though.  I
suspect there's a strong hardware component to this, since every BIOS
chip made (since the introduction of 1.2M drives) has had support for
this mode.

As for spindle speed and data rate, the drives were designed for this
capability.  The software has merely to program the controller and drive
through well-known I/O ports.

-- David

 
 
 

Floppy format problem.

Post by Charles Sulliva » Wed, 26 Apr 2000 04:00:00



Quote:> Charles Sullivan writes:
> > <Regarding formatting a 360K disk on a 1.2M drive.>
> > It turns out that using fd0h360 (as suggested by David C) instead
> > of fd0d360 works, and if I then use:
> >   mkfs -t msdos /dev/fd0h360
> > I get a diskette which is compatible with DOS/Win 98.

> >      the usual 1.2 Meg drives are designed to at least read 360K
> > double density floppies, and I was able to successfully format one
> > under DOS/Win 98 (dual boot on the same box).  The drives change speed
> > and data rate depending on the density.

> > What is undoubtedly true is that this diskette will be unreadable in a
> > standard 360K double density drive with its wider heads.  I think my
> > 10 year-old drive, a Teac FD55GFR-142, had some sort of dual heads
> > which allowed writing either wide or narrow tracks, but I can't find
> > the specs for it on Teac's website.  (I'm pretty sure I used 360K
> > floppies to move files back and forth between one of the antiques I
> > used to play with and the PC with the Teac -142.)  The specs for the
> > new one, FD55GFR-7193, don't mention this.  I trashed my last antique
> > with a double density 5.25" drive about 5 years ago so can't check it
> > out.

> If I remember correctly from my DOS days....

> A 360K disk in a 1.2M drive is written by double-writing the tracks.
> That is, the head writes two adjacent identical tracks for each track
> written by the software.  This way, the result is wide enough for a 360K
> drive to read.

> The effect isn't perfect, however, and some 360K drives do have problems
> reading disks that are formatted in 1.2M drives, but it works most of
> the time.  (I never had a problem with 360K drives reading disks
> formatted in my Teac 1.2M drive, but I have had problems with 360K disks
> formatted in an old IBM PC/AT.)

> I'm not sure if all this is done by the drive and controller (after
> receiving some kind of mode-setting command) or by software, though.  I
> suspect there's a strong hardware component to this, since every BIOS
> chip made (since the introduction of 1.2M drives) has had support for
> this mode.

> As for spindle speed and data rate, the drives were designed for this
> capability.  The software has merely to program the controller and drive
> through well-known I/O ports.

> -- David

You are most likely correct about this.  I went rooting through my archives
and found the data sheet on my old Teac 1.2M drive - it says the same thing
as the specs for the newer one, i.e., 360K is read-only.  I also found the
old IBM-AT Users Guide which says that a 360K diskette written with the 1.2M
drive will thereafter only be readable on the 1.2M drive.  So it's probable
that the (later) BIOSes (or DOS?) solved the problem by writing a double
track - the FD controller looks like a pretty dumb card in itself.

It'd be interesting to determine whether the Linux FD driver does the same
thing.  It's academic for me at this point as I no longer have a 360K-only
drive, but it's nice to know what's going on.

Regards,
Charles Sullivan

 
 
 

Floppy format problem.

Post by Gene Hesket » Wed, 26 Apr 2000 04:00:00


Unrot13 this;

Gene Heskett sends Greetings to Charles Sullivan;
[huge snip]

 CS> You are most likely correct about this.  I went rooting through my
 CS> archives and found the data sheet on my old Teac 1.2M drive - it says the
 CS> same thing as the specs for the newer one, i.e., 360K is read-only.  I
 CS> also found the old IBM-AT Users Guide which says that a 360K diskette
 CS> written with the 1.2M drive will thereafter only be readable on the 1.2M
 CS> drive.

I just spent a few minutes rummaging here, but that binder has
apparently gone to long term storage in the ba*t.

  So it's probable that the (later) BIOSes (or DOS?) solved the
 CS> problem by writing a double track - the FD controller looks like a pretty
 CS> dumb card in itself.

Bad idea, see my reply to a David C. for why.

 CS> It'd be interesting to determine whether the Linux FD driver does the
 CS> same thing.  It's academic for me at this point as I no longer have a
 CS> 360K-only drive, but it's nice to know what's going on.

I'm in that same boat, my linux box has only a 1.44 3.5" floppy, and of
course an ls120 which can also do those.

Cheers, Gene
--



                        email gene underscore heskett at iolinc dot net
This messages reply content, but not any previously quoted material, is
? 2000 by Gene Heskett, all rights reserved.
--

 
 
 

Floppy format problem.

Post by Gene Hesket » Wed, 26 Apr 2000 04:00:00


Unrot13 this;

Gene Heskett sends Greetings to David C.;

Lemme do some creative snipping here...

Quote:>>> Overall, trying to make a 360k diskette in a 1.2 megger will have to
>>> deal with the fact that its probably can only be read in that same
>>> drive, no real 360k drive, with its wider heads, would ever give you
>>> an error free read.

>>> Did I mention its a bad idea?  The only snowballs chance of making it
>>> work is if you started with a bulk erased diskette, one erased by one
>>> of those $30 tape erasers they sell at radio shack.  You could
>>> probably put that disk in a real 360k drive and read it, but if you
>>> ever wrote to it, and then tried to re-write that same area again in
>>> the QD drive, at least that file would be trashed forever.

>> It turns out that using fd0h360 (as suggested by David C) instead
>> of fd0d360 works, and if I then use:
>>   mkfs -t msdos /dev/fd0h360
>> I get a diskette which is compatible with DOS/Win 98.

If its bulk erased first I'd think...

Quote:>> What you've dscribed out would have merit were it not for the fact
>> that the usual 1.2 Meg drives are designed to at least read 360K
>> double density floppies, and I was able to successfully format one
>> under DOS/Win 98 (dual boot on the same box).  The drives change speed
>> and data rate depending on the density.

>> What is undoubtedly true is that this diskette will be unreadable in a
>> standard 360K double density drive with its wider heads.  I think my
>> 10 year-old drive, a Teac FD55GFR-142, had some sort of dual heads

Humm, lemme see if I have any data on the FGR-142, I've got quite a
binder full of teac stuffs on the top shelf.  Ooops, darn, I don't spot
it immediately, it may have gone to the ba*t with another computer I
unplugged the last time about a year ago.  I know I have data on the F
and on the G, but wasn't aware they ever made what sounds like a
composite of the two.  Dual heads? Interesting.

Quote:>> which allowed writing either wide or narrow tracks, but I can't find
>> the specs for it on Teac's website.  (I'm pretty sure I used 360K
>> floppies to move files back and forth between one of the antiques I
>> used to play with and the PC with the Teac -142.)  The specs for the
>> new one, FD55GFR-7193, don't mention this.  I trashed my last antique
>> with a double density 5.25" drive about 5 years ago so can't check it
>> out.

I know the feeling well except mine are in the ba*t, and could be
brought back to life in less than an hour.  But don't count the
excavation time in that hour please.

 DC> If I remember correctly from my DOS days....

 DC> A 360K disk in a 1.2M drive is written by double-writing the tracks.
 DC> That is, the head writes two adjacent identical tracks for each track
 DC> written by the software.  This way, the result is wide enough for a 360K
 DC> drive to read.

It might be a method to experiment with, but in terms of writing 2
identical adjacent tracks such that they look like one to the wider
head, is a once in a blue moon success story, and can't be any other
when you throw in the inevitable wibbles in the motor speed which would
cause severe phasing errors between the 2 tracks when applied out in the
real world.  Those drives often had a bit of slippage in the disk
clamping, and such unwanted slippage is the primary reason the 3.5"
drives pin drive the disk.  At least the disks speed wibbles are
synchronized when the disk is re-inserted, so the only thing that
changes is the physical wavelength of the magnetic domain on the media.

 DC> The effect isn't perfect, however, and some 360K drives do have problems
 DC> reading disks that are formatted in 1.2M drives, but it works most of
 DC> the time.  (I never had a problem with 360K drives reading disks
 DC> formatted in my Teac 1.2M drive, but I have had problems with 360K disks
 DC> formatted in an old IBM PC/AT.)

 DC> I'm not sure if all this is done by the drive and controller (after
 DC> receiving some kind of mode-setting command) or by software, though.  I
 DC> suspect there's a strong hardware component to this, since every BIOS
 DC> chip made (since the introduction of 1.2M drives) has had support for
 DC> this mode.

Yes, its hardware.  IIRC the use of lines that were (on other platforms)
used for disk ready lines became sufficiently cross-dressed to be used
as hardware control switches.  The one impediment to swapping drives
around after the plug styles are compensated for.

 DC> As for spindle speed and data rate, the drives were designed for this
 DC> capability.  The software has merely to program the controller and drive
 DC> through well-known I/O ports.

Later drives were spindle speed and write currant controlled.  The data
rate is the exclusive domain of the FDC chip itself.  All of the ones
I've had from the old Mitsubishi 4853 (what a piece if *that was) on
up were all configured via flea jumpers only.  The 4853 was so poorly
shielded it couldn't run in the same room with a color monitor.

Cheers, Gene
--



                        email gene underscore heskett at iolinc dot net
This messages reply content, but not any previously quoted material, is
? 2000 by Gene Heskett, all rights reserved.
--

 
 
 

Floppy format problem.

Post by Harold Bowe » Thu, 27 Apr 2000 04:00:00





> > Charles Sullivan writes:
> > > <Regarding formatting a 360K disk on a 1.2M drive.>
> > > It turns out that using fd0h360 (as suggested by David C) instead
> > > of fd0d360 works, and if I then use:
> > >   mkfs -t msdos /dev/fd0h360
> > > I get a diskette which is compatible with DOS/Win 98.
 [snip]
> You are most likely correct about this.  I went rooting through my archives
> and found the data sheet on my old Teac 1.2M drive - it says the same thing
> as the specs for the newer one, i.e., 360K is read-only.  I also found the
> old IBM-AT Users Guide which says that a 360K diskette written with the 1.2M
> drive will thereafter only be readable on the 1.2M drive.  So it's probable
> that the (later) BIOSes (or DOS?) solved the problem by writing a double
> track - the FD controller looks like a pretty dumb card in itself.

> It'd be interesting to determine whether the Linux FD driver does the same
> thing.  It's academic for me at this point as I no longer have a 360K-only
> drive, but it's nice to know what's going on.

> Regards,
> Charles Sullivan

The 1.2M drives step twice for each track when mounted as /dev/fd0h360.
I use this mode to transfer files from some of my 8-bit CP/M systems
operating a program which allows them to read and write DOS-formatted
360k diskettes.  The problem with interchangeability relates to the fact
that 1.2M (and dos 720k drives) have twice as many tracks as a true 360k
drive, so the best (and only reliable) way to make the procedure work is
bulk-erase the diskettes, then format them in a 720 or 1.2M drive as
360k diskettes, then NEVER write to the disk in a true 360k drive,
otherwise the latter write will smear a track twice as wide as the
higher drives, and any subsequent writes to that sector will, at best,
be erratic.

Hal

 
 
 

Floppy format problem.

Post by Charles Sulliva » Sat, 29 Apr 2000 04:00:00





> > Charles Sullivan writes:
> > > <Regarding formatting a 360K disk on a 1.2M drive.>
> > > It turns out that using fd0h360 (as suggested by David C) instead
> > > of fd0d360 works, and if I then use:
> > >   mkfs -t msdos /dev/fd0h360
> > > I get a diskette which is compatible with DOS/Win 98.

> > >      the usual 1.2 Meg drives are designed to at least read 360K
> > > double density floppies, and I was able to successfully format one
> > > under DOS/Win 98 (dual boot on the same box).  The drives change speed
> > > and data rate depending on the density.

> > > What is undoubtedly true is that this diskette will be unreadable in a
> > > standard 360K double density drive with its wider heads.  I think my
> > > 10 year-old drive, a Teac FD55GFR-142, had some sort of dual heads
> > > which allowed writing either wide or narrow tracks, but I can't find
> > > the specs for it on Teac's website.  (I'm pretty sure I used 360K
> > > floppies to move files back and forth between one of the antiques I
> > > used to play with and the PC with the Teac -142.)  The specs for the
> > > new one, FD55GFR-7193, don't mention this.  I trashed my last antique
> > > with a double density 5.25" drive about 5 years ago so can't check it
> > > out.

> > If I remember correctly from my DOS days....

> > A 360K disk in a 1.2M drive is written by double-writing the tracks.
> > That is, the head writes two adjacent identical tracks for each track
> > written by the software.  This way, the result is wide enough for a 360K
> > drive to read.

> > The effect isn't perfect, however, and some 360K drives do have problems
> > reading disks that are formatted in 1.2M drives, but it works most of
> > the time.  (I never had a problem with 360K drives reading disks
> > formatted in my Teac 1.2M drive, but I have had problems with 360K disks
> > formatted in an old IBM PC/AT.)

> > I'm not sure if all this is done by the drive and controller (after
> > receiving some kind of mode-setting command) or by software, though.  I
> > suspect there's a strong hardware component to this, since every BIOS
> > chip made (since the introduction of 1.2M drives) has had support for
> > this mode.

> > As for spindle speed and data rate, the drives were designed for this
> > capability.  The software has merely to program the controller and drive
> > through well-known I/O ports.

> > -- David

> You are most likely correct about this.  I went rooting through my
archives
> and found the data sheet on my old Teac 1.2M drive - it says the same
thing
> as the specs for the newer one, i.e., 360K is read-only.  I also found the
> old IBM-AT Users Guide which says that a 360K diskette written with the
1.2M
> drive will thereafter only be readable on the 1.2M drive.  So it's
probable
> that the (later) BIOSes (or DOS?) solved the problem by writing a double
> track - the FD controller looks like a pretty dumb card in itself.

> It'd be interesting to determine whether the Linux FD driver does the same
> thing.  It's academic for me at this point as I no longer have a 360K-only
> drive, but it's nice to know what's going on.

Uh, oh...perhaps there's more to this than meets the eye.  I timed the
formatting
of 1.2M and 360K diskettes on the same high-density Teac drive
(FD55GFR-7193-U)
under Linux (RH 6.0) and DOS (with Win 98) on a Pentium II - 450 MHz system.
         Linux  DOS
  1.2M    89 s  68 s
  360K    46 s  33 s

With the factory jumper settings, the drive speed is supposed to be 360 rpm
for both high and double density.  If the tracks on the 360K were written
two-by-two with the same data to emulate the wide head of a double-density
drive, one would expect the times to be approximately the same, but it
looks like only every other track was written.  (The shorter times under
DOS are probably attributable to it verifying each track immediately after
writing, whereas Linux makes a separate verification pass.)

I also checked the jumper settings on my 10 year-old Teac FD55GFR-142-U
and they are the same factory settings, i.e., double-density speed is
360 rpm.  This drive is installed on an old 33 MHz 486 system running
DOS 5.0 and the times for formatting 1.2M and 360K diskettes are
approximately the same as shown above.

The test diskettes were not bulk erased and the 360K diskette had been
previously written on a double-density drive.

Now I'm more confused than ever.

Regards,
Charles Sullivan

 
 
 

Floppy format problem.

Post by Jean-Yves Sim » Sat, 29 Apr 2000 04:00:00


And thanks also from me. In the past, it was never possible to format
a 1.4M floppy on my notebook. I used /deb/fd0h1440 and got the same
errors as Mr Sullivan posted. I read all the books, posted here. The
problem was never solved. Because of this thread, I looked again
and "discovered" that an /dev/fd0u1440 existed. I tried it...
and it worked !!!

Thanks !


: >
: > Has anyone successfully formatted a 5.25" double-density 360K floppy
: > diskette on a high-density 1.2 Meg floppy drive under Linux,
: > specifically RedHat 6.0?  Or any version of Linux?
: >
: > I use the command:
: >   fdformat /dev/fd0d360

: use /dev/fd0h360

:         d360 is for 360K drives
:         h360 is for 360K disks in 1.2M drives

: man fd(4) for more information.

: -- David

--
 Jean-Yves SIMON       Tokyo, Japan

 
 
 

Floppy format problem.

Post by Harold Bowe » Sat, 29 Apr 2000 04:00:00



  [snip]
> Uh, oh...perhaps there's more to this than meets the eye.  I timed the
> formatting
> of 1.2M and 360K diskettes on the same high-density Teac drive
> (FD55GFR-7193-U)
> under Linux (RH 6.0) and DOS (with Win 98) on a Pentium II - 450 MHz system.
>          Linux  DOS
>   1.2M    89 s  68 s
>   360K    46 s  33 s

> With the factory jumper settings, the drive speed is supposed to be 360 rpm
> for both high and double density.  If the tracks on the 360K were written
> two-by-two with the same data to emulate the wide head of a double-density
> drive, one would expect the times to be approximately the same, but it
> looks like only every other track was written.  (The shorter times under
> DOS are probably attributable to it verifying each track immediately after
> writing, whereas Linux makes a separate verification pass.)

There are two ways of having 1.2M drives work at the 360K density, but
most newer drives do not include the options for one.  The original way
was to actually change the spindle speed; 360 rpm for 1.2MB HD, 300 rpm
for 360/720K densities.  That way, the data rates could be held at the
500 kbps for 'HD' and 250 kbps for 'MFM' (or Double-Density).  This
complicated bios routines because switching speeds required insertion of
considerable delays normally associated with spinup times.  Another
complication introduced with the 3.5" HD format was that the polarity of
the "HD" lead was the opposite between the two.  This is well documented
in older National Semiconductor (among other) data sheets.  This is NOT
how most clone PCs work.

Most of them now keep the spindle speed at 360 rpm and change the data
rate from 500 kbps for 1.2M HD to 300 kbps for 360/720K densities.  If
you work out the math, 300 kbps at 360 rpm is the same bit density as
250 kbps at 300 rpm, so the disks were compatible in this regard.  Now a
density change no longer need accomodate the physical delays in changing
motor speed.

Quote:> I also checked the jumper settings on my 10 year-old Teac FD55GFR-142-U
> and they are the same factory settings, i.e., double-density speed is
> 360 rpm.  This drive is installed on an old 33 MHz 486 system running
> DOS 5.0 and the times for formatting 1.2M and 360K diskettes are
> approximately the same as shown above.

> The test diskettes were not bulk erased and the 360K diskette had been
> previously written on a double-density drive.

I believe that you may be confused by thinking that the 360K diskettes
are writted in a 1.2M drive by writing two adjacent tracks with the same
data...they are NOT.  In a 1.2M drive, the drive steps twice for each
track increment, writing every other track.  If it were not so, your
times for Linux would be more than twice those for DOS.  If you sketch a
couple of tracks on a piece of paper, label one on an end as track 0,
and increment numbers for the adjacent 'tracks'.  Mark a large swath
around track 0, 2, 4, etc.  That would correspond to tracks 0, 1 and 2
on 360K.  The narrower tracks will be reliable, but writing narrow and
reading wide (with a 'real' 360k drive) might not if the disk were
previously written with a 'real' 360k.

If you can find some old 720K drives by TEAC, you will find a jumper
that actually forces a double-step the drive.  It appears that they used
a common stepper and electronics between 40 and 80 track drives (at
least the ones I had in the mid 80s), and the technology was carried
forward.  You can find assembly code for Z80/8080 systems which do the
double-stepping in CP/M archives as part of BIOS code for CP/M.

Hal

- Show quoted text -

Quote:> Now I'm more confused than ever.

> Regards,
> Charles Sullivan