On Mon, 8/3/2026 11:12 AM,
crasso@nycap.rr.com wrote:
I think my old 2008 homebuilt Abit pc is finally coming to the end of
its journey. It has started hanging on the boot at the truck picture
logo (where you have the option of Delete-ing into the BIOS). I could
press Enter or Tab a few times and it would stagger on, get to my BCD
menu and then boot successfully any OS I wanted.
I replaced the CMOS battery and all was well - for a day, next AM it
hung again, first boot of the day. The old battery measured 3.2v
after I removed it. The new one had measured 3.3v before I put it in.
I don't think I cleared the BIOS, just stuck the new battery in.
Anyone, (Paul?) have any ideas how to save my precious old sweetheart?
The two motherboards here that failed, it looked like maybe an
onboard regulator for a chipset voltage, took them out.
On one of them, I somehow figured out that "sleep" wasn't
good for it. Two months went by, I didn't sleep it, it
was running OK. Then I forgot, selected sleep... and it
didn't come back and it would not POST either. I'd killed
it, through stupidity.
If the designer of it, was in the room, there's not
even a guarantee the designer can debug it. One thing
that helps, is to use the DFI (Diamond Flower) technique
of soldering scope-points to the board, and, um, labeling
them. Then you can check all the voltages on the Southbridge
and see if anything there is out of spec. No other brand
really adds anything useful for monitoring in that way.
There is an extensive back-feed network on the board. These
are custom designed, and consist of bailing wire and binder
twine. The purpose of this, is to prevent current flows
from one subsystem, keeping another subsystem awake and
preventing it from "cycling" properly. The board has
to be able to actuate its reset circuit just before POST
and so on. The RTC and CMOS RAM area of the Southbridge,
sit in a "well" powered by the battery, and transmission
gates are used in series with signals, to prevent
current flows out of the well and into other stuff.
Those are some examples of squidgy parts of boards.
I have trouble even reading the logic in the back-flow
prevention circuits. I'd need to go node by node, to help
me figure out how the damn thing works.
The BIOS chip can become flaky. Maybe it can be re-flashed.
Parts of the BIOS are actually write-able by the BIOS,
like the DMI/ESCD, and while NOR flash can take a lot of
writes, maybe at some point their lifespan is exceeded.
The main BIOS module is not written all that many times.
So it can be a candidate for being re-flashed.
But a lot of times, the problem is something at a low
level. With digital logic, it's almost never an internal
gate. The problems are I/O pads and the copper wires.
Copper PCBs are protected by passivation, like HASL
(Hot Air Solder Leveling), to put a layer of bulletproof
metal over the copper. Computer boards are half ounce copper
on signal layers, and the boards really aren't all that good
of a quality. The boards at work, they were good enough
to "stop a bullet". It was hard to lift a foil on a board
at work. Foils come off computer boards, with hardly
any effort at all (I changed out a voltage regulator
on the 440BX to make it Tualatin-compatible and lifted
a foil AND I WAS BEING CAREFUL!). Not one of my finer
moments, but the foil I lifted didn't do anything
so lifting it was not a problem. I just hate bad
workmanship like that.
If you're in the market for a new motherboard, remember
that component size has dropped. They put 0402 and 0201
on these boards. Even a tiny "bump" of a PCB while you
are sliding it into place, can tear an R or a C off the
board (these are resistors and ceramic caps near the
edge of the board). This means, when selecting a computer case
to accept a modern board, there can't be ANY barriers in
the way. Like that big square bar that goes across
the case, that's going to cause handling problems as you
try to drop the board (gently) into place.
My first Windows PC build, it had the right idea. I don't
even know the brand of computer case. The outside skin was
just awful (the machine hasn't had the skin on it for
decades now). But the inside, they were geniuses.
There is no bar across the middle of the case. The
lower bays are small enough, you can put a full length
PCIe video card in there (and the designers of that case,
didn't even know how long video cards would eventually become).
The motherboard tray removes at the back, which allows you
to fit the PCB onto the tray, rotate the tray into place,
do up the screws on the back. The case does a lot of the
right things. But the outside of it is a disaster, and that's
where all the clueless mistakes are, is on the outside.
An old fashioned TV repair man, could likely take a stab at
your Abit, IF he had a schematic. People in the repair business,
they can figure out some amount of how it works. But if you
were debugging the backfeed protection (which would make
for "flaky" behaviors), you really really need a schematic
for that. It's just too hard to trace signals for that,
and make sense of the responses.
Some failures can be caused by dirt and moisture
near high impedance circuits. I'll leave it to you
to decide if the case is super clean, or it looks
like one of my cases. It's unlikely to be a tin whisker,
as tin/lead alloy doesn't grow whiskers. It generally
takes pure tin to do that.
Checking the voltages is where I would start, but this
is only practical if someone marked them and left pads
for them. Diamond Flower soldered scope-points onto their
pads, and that was the pinnacle of assistance to debug,
was having a feature like that. But the cost-reduction
staff at the company, won't put up with that, and the
next board, those have to come off.
If the Southbridge was an ICH4, those can definitely
blow out (latchup/USB). The ICH5 (I lost one), may have a
similar problem but we never got even one web page
confirming that. One motherboard company was brave
enough to warn us about ICH4. Southbridges use
five different voltage supplies, and bringing those
up in the wrong sequence, can damage your SB/PCH.
So not only must the onboard voltage regulators perform their
function, their temporal behavior also matters. To be watching
stuff like this, would take a four channel scope (which
was standard on a bench at work).
There are plenty of fault modes, a shitload of signals.
As an engineer, I always marvel at how long these things
run and that they don't fail sooner.
Paul
--- PyGate Linux v1.5.19
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)