From: "Jules Richardson"
<julesrichardsonuk at yahoo.co.uk>
On Thu, 2005-03-10 at 17:18 -0800, Steven Canning wrote:
Jules,
I proposed doing exactly that (i.e. passing "raw" floppy data through a
parallel port) last year along with some simple math that showed it could be
done.
Yep, I hadn't forgotten :-) (I made sure I kept all the relevant
messages too)
Wasn't the problem that the parallel port just doesn't have the raw
speed for it to work for all disks though? At which point buffering
becomes necessary, which was also the point that everyone went quiet on
the subject :-)
With the cable you mention, how are you supposed to get the created disk
image back onto a disk again? (Personally I need to be able to restore
data from an image back to a floppy)
I need to do some reading up on what the floppy drive's write gate
signal does. If spitting data from an image down the 'write' signal wire
at the same speed as it was read is good enough (and 'write gate' is
actually a 'R/-W' signal) then maybe it isn't too complex. However if
write gate is actually dependant on the data stream too (needing it to
be understood) then it could be rather tricky.
Hi
There are two write signals. One is write data and the other
is the write gate. The write gate is like a write enable. It
also turns on the erase signal the cleans the edges of the track
that is being erased and over written.
The drive has nothing to set the timing of the signals coming
to it. You can not just directly use the recorded data without
realigning the clock edges. This includes the possibility of
write compensation as well.
There are rules about when you use the compensation. It helps to
understand a little about digital recording. The write signal
just makes a bunch of static +- signal levels on the disk. When
played back, the head only sees the transitions and not the levels.
This means that the effective edge of the transition is what is
important. When the signals are on the surface of the disk,
they are a bunch of magnetic fields. They interact in such a way
that there is an averaging or frequency loss if too many transitions
are close together. Compensation helps to avoid this. It is
best to put some compensation in both the receive an transmit
ends. It is simpler is to just use it in the transmit ( or write )
circuit. This is what is done for writing higher density floppies.
If you record all the info played back, you need to realign
the timing and possibly add compensation. It is not an impossible
task but does require some knowledge of the particular technology
used for that floppy format.
You also don't need to know what the data is, just what the
rules are for placing levels on the disk. It is even conceivable
that one can do a better job than the original controller did.
One can do some test signals on that particular media and
determine just how much compensation works best on each track!
Like the 6 million dollar man, "we can rebuild you better than
before".
Dwight
Note that personally I *don't* need to understand the image on the host
machine (at this stage) - all I care about is backing up floppies to
modern hard disk and being able to recreate them again. Understanding
via software decoding might be nice one day, but in the shorter term my
concern's with all the thousands of disks we have at the museum with
data on that are likely decaying...
cheers
Jules