Another disk imaging project
Dwight K. Elvey
dwight.elvey at amd.com
Wed Aug 3 12:15:03 CDT 2005
>From: "Patrick Finnegan" <pat at computer-refuge.org>
>Not to discourage you, but, I'd echo the sentiment above. I'd suggest
>that if you do want to produce something, make something that can read
>the disk on a flux-transition level, and store that sort of data.
>Something nice and generic. That way, you can do all the processing in
>software, and handle any format of disk that comes your way.
I've always had in the back of my mind, the thought of
building something to capture the raw flux-transitions.
I still think the easiest way is to use the SPI port of
a DSP, like the 2181 or similar. These processors are
quite fast and can post process the raw data into something
that can be transferred to the host processor in a more
condensed and slower rate.
To keep Ethan happy, he could even used one of the old
modem cards made by Digicom and Cardinal that have 2115
and 2111 processors on them. A few hardware hacks and
it could be a disk data processor. These two modems,
called soft modems can have the entire software loaded
from a PC.
For those that don't have one of these older boards, a
2181 development board might be the way to go. It makes
sense to keep the back disk controls from something simple
like the parallel port and just use the DSP for the
read/write of data.
The advantage of such a system is that the part count is
minimal. The 2181 has all of its RAM inside. It can bootstrap
from a 8 bit EPROM or FLASH. One can connect up a simple
parallel or serial to a host PC. It can sample really fast.
For those that really need it, it can be programmed in "C"
( although the assembly is very clean ). The entire unit
can be done in less than 10 IC's.
There are distinct advantages to keeping the data processor
separate from the host. One doesn't need to be concerned
with how to handle other system level interrupts. One
can buffer the data for safer transfer rates to the
host system's mass storage.
One thought, although the CatWeasel used sample rates on
the order of 40MHz, adding a simple flopflop to the input
significantly reduces the sampling rate needed( as I've
stated before ).
More information about the cctech