processor voltage mod for the P4T-E motherboard

Disclaimer: this mod will be extremely risky if you don't have a soldering iron with a very small tip and some magnification so you can see what you're doing. You can destroy your mb and/or processor if you're not careful. That said, on to the mod....


I've sucessfully modded my P4T-E for increased processor voltage. The ADP3165 voltage regulator feedback pin is pin 9. This pin has a capacitor to ground and a direct connection to vcore. If the pin connected to vcore through a resistor the mod would be simple, just connect a resistor of the correct value from pin 9 to ground. Since pin 9 connects directly to vcore a trace must be carefully cut. The pin 9 connection to vcore comes from a feedthrough and short pc trace to the capacitor and then to pin 9. Carefully cut the trace between the feedthrough and capacitor (don't cut too deeply into the board, the trace cuts easily). Now you'll need to install a couple of resistors to drop the voltage from vcore to pin 9. The easiest place to install these is by the cpu socket (just be carefull that nothing sticks up so high that it shorts to your cpu heatsink). There's a couple of spots where bypass capacitors could be installed but aren't, these make great vcore and ground connect points. (You'll need a larger soldering iron tip for these connections than what you should use to solder to pin 9). Call the resistor that connects to vcore R1 and the one that connects to ground R2. These should be connected together so you get R1 to vcore on one end, R1 connecting to R2 and then the other end of R2 connected to ground. Use a piece of small gauge insulated wire (solid rather than stranded preferred) to connect pin 9 to the R1/R2 connect point. It's best to keep the wire down on the pcb to minimize noise pickup. Solder the wire directly to pin 9, but carefull not to short 9 to pins 8 or 10.

The voltage will increase by a factor (1+R1/R2). I used 100 ohms for R1 and 1800 ohms for R2. This gave me a voltage increase of 5.6%, which was all the extra I needed to be able to run my 2.2 NW stable at 133mhz fsb. Note that the ADP3165 has over and under voltage protection set at +/-20%. If you don't do the VID pins wire trick, your max voltage will be 1.5x1.2=1.80v if you use jumper mode, or 1.65x1.2=1.98v if you use jumper free mode. If you wire VID5 and VID4 pins together on the processor you can get 1.70x1.2=2.04v max before overvoltage protection kicks in. Note that the exact values of R1 and R2 aren't important, just their ratio. If you use a potentiometer for R2 so the voltage can be easily varied, I'd suggest putting a fixed resistor in series with it to set a lower limit for R2. You could also use a pot for R1 instead, just be carefull of the noise pickup on the R1 wiring, which is more important than R2 wiring. A 200 ohm pot for R1 and 1800 ohm resistor for R2 would give a voltage increase of 0% to 11%.

Final disclaimer, I'm not responsible if you screw up your motherboard or processor. Be careful!

Thanks BMG but I don't think I stand alone when I read your post: Duhhh.

Seems rather complicated. And if any soldering is required, that leaves most of us out in the cold. From what you say, don't think you can use those Pomona Grabbers.

'Guess I just have to wrap the pins.

BTW, are you the BMG on Anandtech? If so, did you ever find out if the newer Samsung RDRAM hits 150 FSB? Somebody on the Anandtech site feels all RDRAM will hit 150, it's just the Northwood that is the limiting factor.

Yeh I'm sometimes on Anandtech also.

Unfortunately I don't think the P4T-E can be modded for increased voltage without doing pretty much what I outlined. If ASUS had put a resistor between the voltage regulator feedback pin and the processor voltage a simple mod would have been possible. If the mod had been easier I suspect it would have been done before now. I've got access to good equipment at work, so decided to give it a try. Fortunately it worked fine.

On the Samsung rdram, I tested quite a number of parts and found that the 128mb with 8 devices all did at least 133 for me. I had some old parts from early 2001 that did 148mhz. I got some 128mb/8 device parts from Googlegear with an early 2002 date code, and 3 did 147 and the other 1 barely did 133. I tested 8 of the 256mb/16 device parts, and all would do 133, but 4 would just barely make it and the other 4 were good to 138 solid with turbo enabled. I tested 4 of the newer 256mb/single sided/8 device parts and none would go over 125.

I couldn't get any parts to 150, but that might be my motherboard or maybe I need to bump the rdram voltage up a bit. I'll probably bump the rdram voltage up a couple of tenths and see if I can get any of my good 128mb parts above 150.

It's interesting what you found out about the RDRAM. Cold Disposition, who posts on this forum, got two 128-MB, single-sided sticks from Googlegear recently and he got it to 150.

After reading your post on Anandtech, I was able to find a whole batch of early 0119 128-MB sticks here in Japan (the same time frame of your early sticks), so I bought a couple. If they can get to 148 too I'd be happy.

I'm going to have a thick 120 mm Sankyo Denki fan blowing over the RDRAM to see if it helps.

My P4T-E has the -13 ICS chips (wouldn't want to solder on that board). And I got a Malay 1.6a. But I gotta get some fine wire for that darn vid pin trick. I'm pretty sure I need at least 1.7 volts to get to 2.2.

I got 2496 with my Costa Rica 1.6a without going above 1.65v. You have a really good chance of hitting 2200 without needing to use the wiretrick to boost your voltage.

Hi
I've done that Vid pins mod and driving My 2.0a stable with 20 x 136 , above that I have to change RDRAM settings 3x / turbo disabled. I'm interested about pulling up RDRAM voltage, maybe that's help. BMG you mentioned something about that, do You have mod already done ?I'm using Kingston's 256Mb memories,
chip manufacturer seems to be Samsung and best I've got is 20 x 144 with 4x / turbo enabled.

Hey bmg thanks a LOT for the info. My P4S333-M has an ADP3165 also so I may have to give the mod a try. Let me make shure I'm clear on this.

One one of the empy bypas capacitor connections you connect R1 (vcore) to what would be the + leg of the capacitor and R2 (ground) to what would be the - leg of the capacitor. Then R1 and R2 are connected on the other end connecting to feadback Pin 9 correct? Could you further explain what trace is to be cut? I'm not quite shure what you mean by feedthrough. A picture or a simple diagram would be great.

I would definately be interested in a RDRam voltage mod.

I am running 512MB sticks at 125mhz but PCI craps out any higher... and board won't boot at 133.
I feel a little voltage boost will increase my chances of booting at 133. My 2.0 with wire wrap boots easily at 133 with 3x RDRam multi. so I don't think it will become an issue.

Keep us posted.