16 Jun
2010
16 Jun
'10
3:31 p.m.
Philip Pemberton wrote:
On 16/06/10 22:24, Fred Cisin wrote:
I was going to jump in here too. :)
On Wed, 16 Jun 2010, Philip Pemberton wrote:
I think the issue is getting confused here: this is IN ADDITION to the
time it takes to read the disc. It's currently taking ~45 seconds to scan and
decode 160 tracks (80
track disc, both sides). It'd be nice to get that down by, say, half.
The rotational speed is 300 RPM (unless you modify the drive)
The heads only operate one at a time (unless you modify the drive).
You can only do one cylinder at a time (unless you modify the drive). I'm down to 90 seconds to read an 80-track disc
(which is close to the
theoretical ideal, and certainly "good enough") as magnetic transition
timing data. To convert that timing data to MFM bits, I have to feed it
through the "software data separator" -- which is what's taking 45
seconds per disc.
<sarcasm> hohoho my project is better than yours in some way. </sarcasm>
My amiga floppy project will read a full disk and write an .ADF in
around 50 seconds, total. There's still room for more optimization, too.
You can watch a demo of the current version of project if you want. See
video link a couple posts down or the pictures.
http://techtravels.org/amiga/amigablog/
50s is a pretty reasonable time for me. If I do the time-between-edges
conversion(or do pll) in the FPGA, I can probably chop that time to
around 35-40s. I think I could get it to theoretical maximum.
I transfer the data while I'm receiving it from the drive. So the first
bit gets transmitted to the PC within 10us(!!) of the first pulse being
received from the disk.
Keith