TEMNGT23@ysub.ysu.edu (Lou Anschuetz) (01/30/91)
Ok, folks, I'm really stumped as to how to proceed on this one.
Any and all ideas would really be appreciated.
Background:
machine is a 25MHZ 386 with 64K cache
Orchid Paradise VGA
2 serial, 1 paralled (modem on com3 - works fine)
MFM controller (WD 1003)
ST-225 HD <- I know it's old, but I've used it for 6 years
with never a failure, hate to get rid of it.
3.5 & 5.25 " floppies, both high density
4MB RAM
for purposes of diagnostics, config.sys contains only
buffers and files. No drivers!
autoexec.bat contains only path and prompt $p$g
In any case, machine works flawlessly for roughly an hour. After
an hour, and for at least 72 hours of operation, everything works
works fine unless you try to WRITE to a floppy. Reading the floppies
and/or using something like Norton DT on them is not a problem.
The problem is that when you try to write to a floppy you get,
in 40 column CGA, "Parity Check ????, SYSTEM HALTED"
BTW, the machine will STILL respond to alt-ctrl-delete and
reboot normally. The problem is not as severe at 20MHz as
it is at 25MHz (happens 95% at 25, about 60% of the time at
20MHz). Turn the machine off for 20 minutes and the same
cycle repeats. Take the cover off, and never get any failure -
so clearly this is heat related.
Now, the question. My "best" guess is that either the problem is
a DMA failure, controller failure, or cache chip failure that is
heat related and somewhat speed sensitive. Since all of those
are supposedly tested during POST, one of them fails during
the slow transfer to a floppy (BTW, I have copied entire 5MB
subdirectories on the hard disk without this problem occurring).
1) Am I narrowing it down to the correct circuits?
2) If so, how the heck do I check them? I understand there
is no known way to check SRAM. I do all the controller tests
with Norton's calibrat and all work correctly (but that is
only to the hard disk of course). I can also do a complete
backup to high density 3.5" floppies using PCTools with the
High Speed DMA option and get no failures.
Help! What do I do next (aside from moving to a very cold climate)?
Please write to
Lou Anschuetz
temngt23@ysub.bitnet or temngt23@ysub.ysu.edu
(and tens of other places world wide)