"LarsJ" said:
Hi
I have an ASUS A7V8X motherboard and I have recently bought a CPU Athlon
XP-2400 Mobile 35 W.
I want to run a cool and quiet PC (not overclocking) and therefore I want to
lower VCORE as much as possible.
This CPU needs a VCORE on only 1.35 V. When I test this CPU on my A7V8X I
realized unfortunately that the lowest Vcore I can set is 1.525V. Is there
an easy way to lower VCORE further? I have upgraded BIOS to the latest 1013.
I saw in this newsgroup some months ago how you can increase VCORE with a
resistor between pin 9 on L6917BD and ground. Can I just do the opposite and
connect a resistor between pin 9 and Vcc (12V)?
I want to have a possibility in BIOS to change between 1.2V and 1.4V.
I wonder now if anyone has tested this approach and if so, the value on the
resistor you have chosen. I would be very grateful if anyone has some
thoughts about this subject.
Thanks,
Lars J
I see a couple of possibilities. The solution requiring the least
knowledge, is to pry the five VID signals on the Vcore regulator
chip, then control them with a five position dipswitch connected
to GND. Prying the pins up is necessary to isolate any GPIO signals
that the BIOS is using to drive those pins. (Use solder wick to
remove the solder, and carefully pry the pin up a fraction of an
inch with a dental pick, while heating with the soldering iron
to free it. Too much prying will snap the pin off, or rip the
copper foil, so be careful). If the BIOS has an option to modify
the VID voltage, then there have to be some GPIO pins connected
to the VID lines, to force them, and you don't want to damage
those signals if possible.
If you want to take the analog modder route, then a resistor
connected to a regulated rail voltage higher than the output
voltage should work. The +12V on a power supply will wander
around a good deal (+/- 5%), while some voltages regulated
by the motherboard will be much more stable. Your wiring
must be kept short, to keep the wire from becoming an antenna.
And, somehow, you'll need to figure out just how much resistance
is required. (Search some of the private forums, to see if
someone has already figured this out.)
The thing that makes me nervous about the analog mod route,
is adding the resistor could be affecting the stability of
the control loop. If you can find instructions on a private
forum of someone who has succeeded at undervolting this
board, that is probably better than me guessing at whether
it is going to work properly or not
If you plan on using a potentiometer to dial the resistance,
remember that dialling the device to zero ohms could
destroy something on the motherboard. There should always
be a limiting resistor in series, to provide a measure of
safety.
So, the first part of your mission, is to find a locally
regulated voltage on the board, like the +2.5 used for the
DIMM or a local +3.3V if there is one. (By using 3.3V,
the resistor value used for the voltage boost should be
in the same ballpark as the resistor value needed to cause
the output voltage to drop, so that is another advantage
of a lower voltage like that. The resistor value would be
a lot higher, if you are connecting to +12V or the like.
Like (12V-Vout)/Vout times higher. Evaluating that
equation for (3.3V-1.65V)/1.65V gives a factor of 1X
compared to the "resistor value to GND" mod.)
For that amount of messing around, I think the digital
approach is less trouble. Practice your soldering skills
on something else first. I have a lot of soldering experience
with more robust boards, and I found the Asus board I worked
on, to use thinner copper foil than I am used to. Delamination
of the foil is pretty easy to do. And those tiny pins
cannot be bent through too large an arc, before they
snap off. Enough clearance to slide some mylar film under
the pin as an insulator, is all that is needed.
An Uber-style BIOS would be another way to get there. If
someone has freed up some settings with MODBIN or the like,
then a modified BIOS might open up more options. The A7N8X
BIOS has been hacked, but I don't know if the A7V8X has been
blessed this way or not.
HTH,
Paul
