Paul wrote:
Bob F wrote:
I have a motherboard I acquired cheaply, but which does not even
post on poower on. This was in a case, and had the 4 pin power
connector plugged in 90 degrees wrong so that the pairs of 12V and
GND connections would each receive one 12V and one GND supplied from
the power supply. I suspect, that the internal board connections to
this connector could have fused. What I'm wondering, is if there
would be any possibility of determining if this is so, and if it can
be repaired. If I turn on the board with the large power supply connector
installed, and the 4 pin connector disconnected, I see no
significant voltage on either 12V lead of the motherboard 4 pin
connector, so voltage is not getting through to the 4 pin
connections from the main power connector. The board is an Asus M4A87TD EVO

Any ideas about how to test my hypothesis, or perform repairs?

ATX12V
12V 12A max 1V ~100A
Yellow ---+------------------------ Vcore ------------ CPU_Core
Yellow ---+ |
Black ------+ |
Black ------+------------------------+--------------- GND

12V 12A max
12V yellow ---+------------+---+------+----------------+
12V yellow ---+ | | | |
(24 pin main) fan fan PCI Express PCI Express --
~2A load, video, high end | | Slot
Power Slot Power -- 4.2A load
(6600 card) | | | | -- (non-video card,
much less) GNDS (multi) ---------------+---+------+----------------+
(The connector current rating varies with wire gauge a little bit.
The ATX12V could handle a bit more current than shown in the diagram,
as the pins have "fewer neighbors", which helps with heat. The above
rating on ATX12V is intended to be conservative. I'm not going to
argue if someone digs up the Mini-Fit Jr datasheet and corrects
the above numbers.)

In that diagram, as far as ohmmeter readings on the motherboard
itself are concerned, there should be no DC path from the yellow pair
on ATX12V, versus the yellow pair on the main connector. If you leave
the ATX supply connected to the motherboard, on the ATX side the
rails could be shared. But with the ATX supply completely
disconnected from the motherboard, there should be no path from
ATX12V to 12V1 pair on the main 24 pin connector.

Exception to the rule, is on Biostar-branded motherboards. For
some obscure reason, users have been finding the two 12V
distributions connected together by Biostar. Users have reported
connecting just the main ATX connector, and the processor runs and
system comes up. This is not particularly good for the main connector,
depending on total electrical load. Biostar was also doing this in the AGP
era.
It's hard to tell whether they're still doing it, unless a poster
volunteers the information. Biostar doesn't admit to the practice,
in the user manual.

The Mini-Fit Jr connector family is keyed by the shape of the
connector plastic. And that's intended to warn the user about
the orientation. The ATX12V also has matching latch elements
on both connector bodies, as a "hint" as to orientation of the
two bits. While a person can override the orientation with
enough pounding, usually a person would notice something was
wrong, before forcing it into place. The connector might not
even want to stay there, as the latch cannot engage if the
connector is installed in any (non-preferred) orientation.

So is something blown ? Yeah, could be. Especially if the
rotation is 180 degrees, and the 12V is applied reversed.
If 12V was shorted to GND, maybe that triggered the OC
protection on the ATX supply. If a monster supply was used
(like 12V 60A single output), maybe something nasty could
happen. But I would have expected the ATX wire harness
plastic to start to melt if that happened. While individual
outputs are supposed to have current limiter implementations
(20 amp limit on some wire sets), we can't always be sure what
is there, because the manufacturer usually does not provide
proper detail about what they've done. The product labeling,
if believed, tells you the whole 60A could be burning that
connector. And I don't think we want to believe that. That
isn't safe.

If you want to bypass the ATX12V connection point, you'd want
to find the input point on the VCore circuit. You can see
a typical VCore implementation here on page 11 (this is the
chip used on my P4C800-E Deluxe). You would want to connect
the ATX12V to the left leg of inductor L1 in the upper left
hand corner. Where it says "Vin 12V".

http://web.archive.org/web/200403310...5ADP3180_0.pdf

But I'd do some serious ohmmeter work first, before doing
any more "live fire" tests :-) For example, use the ohmmeter
to ohm from ATX12V motherboard connector yellow, to the
equivalent of the left leg of inductor L1. You can usually
spot the inductor, because it's slightly different than the
matched inductors on the phases. You would also want to
do a check for a short from 12V to GND, using low power
ohms, on the motherboard. To see if there is a DC short
already present in the circuit.

If you look up the VCore regulator used on the motherboard,
you can see what protections it has on output. Usually it
has overcurrent protection. And that's to prevent the
copper planes of the motherboard from becoming so hot,
the PCB FR4 starts to become charred. That's when the
processor is shorted out for some reason.


I checked what I could. The +12V and Gnd connections on the 4 pin connector both
have continuity between the same signal pin pairs, + Gnd goes to Gnd, and the
12V pins have solid connections to a bunch of what are probably regulators of
some sort. So the problem must be elsewhere on the board or the processor. When
powered up, the fans turn on, and stay on. But other than the 3 keyboard leds
going on when the power button is hit, it does nothing including beeping.

Pressing the "MemOK" button cause the DRAM LED to flash regularly, but that's
all it does. No other results are seen. Clearing the CMOS changes nothing. I
tried restarting it with a USB floppy drive containing a bootable floppy with
the latest BIOS on it, and nothing happened. So much for the CrashFree BIOS3
feature.

Unfortunately, I have no other compatable processors or motherboards to test
these ones. I did test the DDR3 ram on another board, and it works fine. So
until I find a way to test the motherboard and processor, I guess I am at a dead
end.