3.5" floppy drive question(s)

[Chuck Guzis]

On 11 Oct 2008 at 20:47, Mr Ian Primus wrote:
 
> > IBM PS/2s did it optically.
> 
> Hehe. Or not at all... I remember using one PS/2 that just assumed
> everything was HD unless told otherwise (or the disk was already
> readable). Didn't even need to drill holes in the disks to 'upgrade'
> them. Of course, these days, I find myself taping over the holes in
> disks, since DD media is hard to find. 

At one time, diskette "punchers" were a popular item at computer swap 
meets.  The price differential between HD and DD media at one point 
was substantial.  I think I paid over $50 for my first box of 10 Fuji 
DSHD 3.5" diskettes.  I found it easier just to stack up a bunch of 
DD diskettes on my drill press and drill a hole through the entire 
stack.
 
> I've made high density 5 1/4" drives into 80 track double density
> (720K) drives by cutting the trace on the board that leads to pin 2,
> and tying it to ground. This forces the drive into DD recording. 

You're leaving something out here.  While that forces the drive into 
DD recording mode, the spindle speed doesn't change unless the drive 
is also jumpered for "dual-speed" 300/360 mode.  So there can be a 
real difference between a "fixed" drive and a genuine 720K DSQD 
drive.
 
> Now, what I wonder, is how does the _controller_ know what kind of
> disk is inserted in the drive? According to the documentation I have,
> the density select line is defined as a unidirectional signal TO the
> drive, not from it. Having not worked as much with 3 1/2" drives, I
> don't know, but I would assume that the drive senses the disk's hole,
> and operates in that data rate and density recording, and the computer
> simply picks up on the data rate and works accordingly. Or, perhaps,
> in 3 1/2" drives, pin 2 is bidirectional, allowing the drive to tell
> the computer the type of disk inserted. I don't know. 

Grab some documentation on early 1.44MB drives.  You could often 
configure them for "auto sense" (the way most recent drives are by 
default) or "remote sense" in which the status of the media sensor is 
fed back to the host and the host selects the density via pin 2.  Or 
do it the way the PS/2 did and ignore the media type completely.

Some drives are smarter than others.  Most 3.5" USB drives will 
automatically sense 1.23MB media and adjust the spindle speek 
accordingly (they do not, however allow formatting of 1.23MB media).  
I've got a "legacy" interface drive from Toshiba that will also do 
the same.  Appropriately configured 3.5" Teac GF and some HF drives 
can use a signal on pin 4 to change spindle speeds.

There's more variation among HD 3.5" drive signals that most people 
realize, particularly in Japanese variants, where conventions used in 
the NEC 9801 held sway.  A bog-standard 1.44MB PC drive will often 
refuse to work in that architecture.  The 9801 used a very sane 
approach to diskettes--the format of an 8" diskette is exactly the 
same as the 5.25" and 3.5" varieties--77 cylinders, 2 sides, 8 
sectors per side, 1024 bytes per sector and 360 RPM.

Cheers,
Chuck