Overclocking

Raise voltage on boards with FC-socket

© www.hardocp.com by Ed Gannaway

Background
To make this happen you will need a Celeron-II system with a mainboard that has adjustable voltage for the CPU. This article is based on SOYO 7VCA and Celeron II 533MHz. The 7VCA bios voltage tweaks only go up to +10% which for this chip means the voltage maxes out at 1.65 volts. Even at this voltage the Celeron would not run stable at 800 MHz. Running the same CPU in an ABIT BM6 with voltage settings for 1.8 volts it ran stable for over an hour while playing Q3. That's were the fun begins:

After reading the Intel spec sheet (http://developer.intel.com/design/celeron/datashts/) on the different pinouts I came upon the logic table (excerpt below) that determines at what voltage the CPU natively uses. These pins (and locations) are the voltage ID [VID] pins VID0 (AL35), VID1 (AM36), VID2 (AL37) and VID3 (AJ37).

VID 3

VID 2

VID 1

VID 0

Vcc [core]

1

0

1

1

1.50 (Default)

1

0

1

0

1.55

1

0

0

1

1.60

1

0

0

0

1.65

0

1

1

1

1.70

0

1

1

0

1.75

0

1

0

1

1.80

0

1

0

0

1.85

0

0

1

1

1.90

0

0

1

0

1.95

0

0

0

1

2.00

0

0

0

0

2.05

1

1

1

1

No Core


NOTES:
0 = Processor pin connected to Vss.
1 = Open on processor.

The logic table shows the state of these pins from 1.5 volts all the way up to 2.05 volts; the Celeron II is factory set for 1.5 volts. This corresponds to only VID2 being internally connected to Vss (AK36). Different combinations cause the motherboard to supply different voltages.
Now here's the beauty of the whole thing - connect VID1 and VID0 (in addition to the already internally connected VID2) to Vss to make the CPU boot at 1.65 volts! This doesn't change the 10% adjustment factor in the bios, but now it STARTS at 1.65 instead of 1.5 volts! At 10% maximum tweak, the CPU now feasts on 1.8 volts of overclocking goodness whereas before the maximum was 1.65 volts...what used to be the max is now the default.

To accomplish this, I thought about soldering jumper wires across the back of the socket370. Not only would this make the change semi-permanent but also it would void the warranty. To solve this problem, I pulled one strand out of an 18 gauge multistranded wire and, using just a couple of turns on each pin, wire wrapped VID0, VID1, and Vss together on the CPU itself being extremely careful not to contact any of the other pins. The benefits of this option are:
1. Only the CPU is changed.
2. The CPU retains the new preset voltage regardless of motherboard used.
3. The motherboard warranty remains intact and I can swap out the CPU with a PIII without having to remember I had modified the motherboard.
4. It is reversible in about three seconds.

The end result is that my 7VCA will now run my Celeron II at 800 MHz rock solid stable!

Why not go higher, you might ask?
Once again refer to the chart. As you can see up top, the internal VID2 connection (highlighted in yellow) is shared in the ranges of 1.5 to 1.65 volts and 1.90 to 2.05 volts. Without some kick-ass cooling, the upper range is probably out of reach unless you like the taste of Celeron popcorn! Anything between 1.70 and 1.85 volts requires VID2 to be disconnected from Vss. I would venture to say that it is impossible to disconnect this pin internally without destroying the whole thing therefore some external method would be needed. I donít currently know of a reliable (heavy emphasis on reliable) way to insulate a pin on the socket format CPUs. You could cut the VID2 pin off of the CPU (bad idea!) or disassemble the socket370 on the motherboard and desolder the corresponding pin. The first method is permanent while the latter is reversible but requires more than just basic soldering skills. Perhaps someone more enterprising than I will find an easier method but for now 1.65 volts at default is the practical limit.