Quite true. I noticed the difference between the monitor chip
readings and my voltmeter. The voltmeter readings looked
to be very close to nominal and quite steady, while the Probe
readings were wandering around.
I still wouldn't ignore the 3.3V warning however. That could
be indicating that your ATX 20 pin power connector is loose.
Turn off and unplug the computer. Unplug the 20 pin connector
and examine it carefully. Look for melted plastic or discolored
metal on the connector pins. These could be signs of thermal
damage.
The original cause could be something as simple as
the connector not being fully seated, which causes the
resistance between the connector and motherboard to be
higher than normal. The pins get hot, and the metal
oxidizes. The plastic melts next to the pin, if they
get hot enough.
There was one model of Asus motherboard, where users noticed
the connectors fit sloppy on all the boards, implying a
quality control problem at the factory making the connectors.
I have not heard of that problem being repeated, and presumably
Asus no longer buys those connectors.
Once the surface finish of the pins is damaged, the
connection will never be the same again. Even if you
somehow had a technology to clean the pins, they would
still degrade further over time. The original finish of
the pins is very important, as the metallurgy (layers of
metal plating) are selected for long life, and if they
are disturbed, all bets are off.
If the finish of the connectors is damaged, the only way to
guarantee a fix, is replace the connectors on both ends.
That would mean soldering a new connector on the motherboard,
and installing a new mating connector on the power supply.
The other possibility, is the connectors are good, but the
power supply is weak. You will know soon enough, when you
get your 500W supply.
I can't state positively, what devices feed off the 3.3V. I
can tell you, that the board does have a number of linear
regulators (designed low dropout style, to minimize the
heat generated), constructed with MOSFETs and quad op amps.
It is possible the 3.3V is used to power the RAM (via 3.3
to 2.5V linear regulator), and then the 2.5V is further
dropped to voltages like 1.8V or 1.5V, for other chip uses.
Actually, I have some measurement data. DDR400 2-2-2-6 memory.
P4C800E-Deluxe. 2.8Ghz Northwood.
1 stick DRAM single channel - sitting idle in BIOS = 3.3V @ 8.1A
2 sticks DRAM single channel - sitting idle in BIOS = 3.3V @ 9.0A
2 sticks DRAM single channel - memtest86 = 3.3V @~10.5A
2 sticks DRAM dual channel - memtest86 = 3.3V @~11.6A
4 sticks DRAM dual channel - memtest86 (PAT on) = 3.3V @~14.4A
~ means approximate, as memtest current jumps around during
the reading and writing of blocks of data.
The 14.4 amps is spread over three connector pins. The pins
are rated for 6 amps each or 18 amps total. There would only
be a problem, if the currents were not shared fairly by the
three connector pins. Which has happened on a Tyan motherboard...
Paul
