Hi !
I replaced my MB with an A7V600 a few months ago. After that the
machine locked up at regularly. I traced it down to low voltage on the
3.3V line. The problem was actually easily fixed by unplugging the
main power connector to the board, giving it a quick clean and
reconnecting it.
Unfortunately this seems to be only a work-around. The problem
re-appeared four weeks later. And although unplugging/replugging the
power connector to the board helped again, it is not a very satisfying
solution.
As the machine worked fine before I replaced the board, I tend to
think that the PSU/connector is not actually the problem (although I
might be wrong).
As the problem only appears at very irregular intervals (one to
several weeks), it is very hard to test and trace down though.
Has anybody heard of a similar problem? Any ideas on how to check
which component could be at fault?
TIA
Stefan
Once a connector of this type degrades, the problems will continue
to occur until the connector is replaced. At least that has been
my experience with stuff like this (other Molex connectors that
carry large currents). Yeah, I've tried pinching the contacts
and stuff like that, but it just doesn't work. It is a surface
finish issue. Due to the design of the ATX style of connector,
it is impossible to visually check the surface finish of the
contacts (at least without pulling the pins out). Replacing
individual pins (on the PSU end) is another possibility, if
you can find them.
If you don't want to spring for a new power supply, you could
find an ATX extender cable, and depending on how tightly it
grips, that might work better than it does now. The part
I don't like about a solution like this, is there is extra
voltage drop due to the new length of cable, and the connectors
themselves.
http://www.bitpartscomputer.com/systemprices01.htm
ATX POWER EXTENDER CABLE $6.75
In a professionally designed power system, a partial solution to
low voltage is via "remote sense". How that works, is a separate
wire is connected to the "center" of the power distribution
plane, and the feedback system allows the power supply to adjust
the output voltage, so the voltage is exactly 3.3V at the load.
Inside the PS itself, the voltage might be 3.5V at the wiring
harness, but the rest of the voltage drop is caused by the cable.
Unfortunately, the people who invented the ATX standard, cut
a few corners, by not having separate pins for remote sense.
An ATX supply does have a 3.3V remote sense wire, but it shares
a pin with a 3.3V supply wire. You'll find a 3.3V supply pin at
the end of the harness, that has two wires fastened into the
same pin. (An Antec Truepower has this same feature, on 3.3, 5.0
and +12V, and allows all three to be adjusted.) In theory, if the
thinner of the two wires that go into the +3.3V pin was connected
to the motherboard, to a 3.3V point in the circuit, then the PS
would be able to compensate for the crappy connector contact.
At least, until the power supply runs out of compliance range
(i.e. it will only raise the voltage and compensate to a point,
before going into overvoltage limiting). Note - this is only
a theoretical observation, and the implementation can be
hazardous to your system if it falls off.
In a financially unconstrained situation, you would replace both
connectors (i.e. that means a new motherboard and a new power
supply, or the services of someone who knows how to replace
them properly - hard to do properly by a hobbyist like me).
This guarantees the problem gets fixed (because we don't know
which end of the connection is at fault). Unsoldering the
motherboard pins is difficult, as the copper around the pins
drains all the heat needed to melt the solder - proper unsoldering
equipment helps. On the PSU end, individual pins can be replaced,
but again, to do it properly, requires redoing all the pins, and
matching the lengths of wires in the cable harness, to prevent
bunching up in the harness.
Depending on your finances, you could try another PS or try
an extender cable. Whether the extender cable will help or not,
depends on whether the improved connector contact resistance is
greater in magnitude than the extra resistance of the length
of wire used.
PSU connector for cable harness - need plastic plus pins
http://dkc3.digikey.com/PDF/T042/0104.pdf
Molex Minifit Jr 20 pin plastic shroud 39-01-2200
Needs female pins (see lower left of page) - 16 guage is fatter
than 18 guage. Pins are difficult to crimp by hand. May need 16
guage pins if stuffing two wires in a pin.
Motherboard connector 20 pin 39-29-9202 (includes pins)
http://www.newark.com/NewarkWebComm...sp?action=0&First=0&QText=39-29-9202&x=26&y=5
Info from molex.com, use the search engine and the above part number.
There are actually many mating connectors.
HTH,
Paul
