5 Feb
2011
5 Feb
'11
9:11 p.m.
Here's one for the PC gurus (I know there are a few of you here!)
I've got the 386 motherboard on the bench. It's got four 1MB 32pin SIMMs
installed (it can take eight), no graphics card and my POST code display
card (which arrived in the post this morning). I've set the jumpers per
the manual:
http://stason.org/TULARC/pc/motherboards/I/INFORMTECH-INTERNATIONAL-INC-486…
The CPU on this thing is an AMD Am386DX-40. The math-coprocessor (Intel
80387DX-33) has been pulled for now. BIOS is an AMI (American
Megatrends) BIOS, label on the chip says Copyright 1992. This is the
BIOS which has the atrocious magenta-and-cyan setup screen.
The POST card is reporting stable power, RESET inactive, and a valid I/O
clock. After powering up, I get the following sequence of POST codes
cur prev
-- -- holds for a second (or so)
02 01 holds for ~15 seconds
06 05 holds for a second
0d 0c holds for a few minutes
00 0d holds forever
The POST card has two displays: CUR and PREV, for Current and Previous
code. CUR is the code most recently written to the debug port, PREV is
the one before that.
Now, according to http://www.postcodemaster.com/AMI91.shtml , 0D is
"CMOS Shutdown Register Test to be Done Next", and 00 is "Going to Give
Control to INT 19H Boot Loader".
Does anyone have any clue what these error codes actually mean? Complete
guess here, but it looks like the BIOS is having trouble talking to the
CMOS RAM chip. Going by the Dallas datasheet, the most likely candidate
is that it's writing 01X (i.e. 0,1,something) to the enable bits in CR1,
then reading it back to make sure the control register was set up
correctly. The readback failed (repeatedly), so it got stuck in a loop.
Obviously without a disassembly of the BIOS, it's pretty hard to say for
definite that this is the problem... but does what I'm saying sound at
least reasonably plausible?
I'm thinking my next step should be to plug a couple of Harwin pin
headers into the turned-pin socket and probe it with the logic analyser...
Does anyone have any other ideas?
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
5 Feb
5 Feb
9:42 p.m.
What did it do when you had a video card and keyboard installed? Was there a
beep code?
The Dallas chip is probably dead (battery), have you tried reworking it with
a coin cell?
http://www.mcamafia.de/mcapage0/jpg/ds1287rw.gif
----- Original Message -----
From: "Philip Pemberton" <classiccmp at philpem.me.uk>
To: "cctalk" <cctalk at classiccmp.org>
Sent: Saturday, February 05, 2011 9:11 PM
Subject: Debugging a 386 motherboard...
Here's one for the PC gurus (I know there are a
few of you here!)
I've got the 386 motherboard on the bench. It's got four 1MB 32pin SIMMs
installed (it can take eight), no graphics card and my POST code display
card (which arrived in the post this morning). I've set the jumpers per
the manual:
http://stason.org/TULARC/pc/motherboards/I/INFORMTECH-INTERNATIONAL-INC-486…
The CPU on this thing is an AMD Am386DX-40. The math-coprocessor (Intel
80387DX-33) has been pulled for now. BIOS is an AMI (American Megatrends)
BIOS, label on the chip says Copyright 1992. This is the BIOS which has
the atrocious magenta-and-cyan setup screen.
The POST card is reporting stable power, RESET inactive, and a valid I/O
clock. After powering up, I get the following sequence of POST codes
cur prev
-- -- holds for a second (or so)
02 01 holds for ~15 seconds
06 05 holds for a second
0d 0c holds for a few minutes
00 0d holds forever
The POST card has two displays: CUR and PREV, for Current and Previous
code. CUR is the code most recently written to the debug port, PREV is the
one before that.
Now, according to http://www.postcodemaster.com/AMI91.shtml , 0D is "CMOS
Shutdown Register Test to be Done Next", and 00 is "Going to Give Control
to INT 19H Boot Loader".
Does anyone have any clue what these error codes actually mean? Complete
guess here, but it looks like the BIOS is having trouble talking to the
CMOS RAM chip. Going by the Dallas datasheet, the most likely candidate is
that it's writing 01X (i.e. 0,1,something) to the enable bits in CR1, then
reading it back to make sure the control register was set up correctly.
The readback failed (repeatedly), so it got stuck in a loop.
Obviously without a disassembly of the BIOS, it's pretty hard to say for
definite that this is the problem... but does what I'm saying sound at
least reasonably plausible?
I'm thinking my next step should be to plug a couple of Harwin pin headers
into the turned-pin socket and probe it with the logic analyser...
Does anyone have any other ideas?
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
10:01 p.m.
On 06/02/11 02:42, Teo Zenios wrote:
What did it do when you had a video card and keyboard
installed? Was
there a beep code?
Black screen, no beeps. I get the same POST code sequence with the VGA
card in as I do without it.
The Dallas chip is probably dead (battery), have you
tried reworking it
with a coin cell?
Not yet -- the chip isn't that old, I'd have thought it would still be
good...
Also, the board packed in after I cleared some corrosion (from an old,
leaky nicad battery) from the PCB and replaced the old dual-wipe RTC
chip socket with an Augat turned-pin socket. Now the board is DOA...
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
11:59 p.m.
Philip Pemberton wrote:
On 06/02/11 02:42, Teo Zenios wrote:
When you removed the old socket did you carefully examine the legs of
the pins to see if you accidentally pulled out any feed through tubes?
What did you do to clean off the corrosion? I hope you used a mild acid
such as white vinegar to neutralize the alkaline (base) leakage then
rinse off with water...
John :-#)#
--
John's Jukes Ltd. 2343 Main St., Vancouver, BC, Canada V5T 3C9
Call (604)872-5757 or Fax 872-2010 (Pinballs, Jukes, VideoGames)
www.flippers.com
"Old pinballers never die, they just flip out"
What did it do when you had a video card and
keyboard installed? Was
there a beep code?
Black screen, no beeps. I get the same POST code sequence with the VGA
card in as I do without it.
The Dallas chip is probably dead (battery), have
you tried reworking it
with a coin cell?
Not yet -- the chip isn't that old, I'd have thought it would still be
good...
Also, the board packed in after I cleared some corrosion (from an old,
leaky nicad battery) from the PCB and replaced the old dual-wipe RTC
chip socket with an Augat turned-pin socket. Now the board is DOA...
10 Feb
10 Feb
1:35 p.m.
On 06/02/11 04:59, John Robertson wrote:
When you removed the old socket did you carefully
examine the legs of
the pins to see if you accidentally pulled out any feed through tubes?
I didn't see any.. although from experience they can be quite small.
I'll see about pulling the socket and continuity-checking between the
top and bottom sides.
Interestingly there are a few areas of the board which aren't the same
brown colour -- there are bits of black around the pins, which look like
they're inside the PCB.
What did you do to clean off the corrosion? I hope you
used a mild acid
such as white vinegar to neutralize the alkaline (base) leakage then
rinse off with water...
Yup. White vinegar brushed onto all the corroded parts, left for a
minute or two, rinsed with water, then a final clean with Chemtronics
Flux-off to drive off any remaining water or grime.
After I soldered the new socket in place, the whole board got a dunk in
isopropyl, including scrubbing all the dirt out from inside and between
the ISA slots, chips, etc. That thing had more dust bunnies than the
Playboy mansion...
--
Phil.
classiccmp at philpem.me.uk
http://www.philpem.me.uk/
4912
days inactive
4916
days old
4 comments
3 participants
participants (3)
-
classiccmp＠philpem.me.uk -
pinball＠telus.net -
teoz＠neo.rr.com