A7N8X SouthBridge Vdd Volt Mod ... The Correct Way

[ViperJohn]

Southbridge VDD Voltage Mod
[UPDATE:- Works on ALL models!]
By MarginJohn

Its a fact that most R1.04/1.06 A7N8X motherboards will have problems with FSB speeds over about 185
to 190Mhz. This is due to the low SouthBridge Vdd voltage (typically about 1.60v) ASUS has the boards set
up for out of the box. You will think its the memory or CPU thats holding you back when in fact its the
motherboard's SouthBridge. If you want to run higher FSB speeds you will probably need to do a SB Vdd
volt mod. Please note that while its called the SouthBridge Vdd mod it raises the Vdd on both the
SouthBridge and the NorthBridge. I called it the SB mod because the SB is right next door to where you
are modding at.

While the Mod discribed in the forums and linked to in the FAQ will work, it is not the correct way to do the it.
This Mod raises the SouthBridge Vdd voltage by effectively creating a voltage divider at Pin 11 Vfb (Voltage
Feed Back) of the CM3708 regulator. Pin 11 Vfb of the regulator is a relatively high impedance, high gain input
which make it very susceptible to noise. Its purpose is to maintain the voltage output of the regulator
that is actually set with a voltage divider (R628/R630 on the MB) at Pin 7 Vref (Voltage Reference) of the
CM3708. For those interested you can download the spec's and reference circuit, which ASUS follows closely,
for the CM3708 at http://www.champion-micro.com/product-en/CM3708.htm

Modding the CM3708 at Pin 11 Vfb can adversly effect the chips ability to regulate and cause instability due to
noise and stray capacitance introduced at Pin 11 Vfb. Due to its high impedance the wire and pot used to
connect to Pin 11 Vfb effectively becomes a damn antenna smack in the middle of RFI hell! Capacitance, even
of the stray kind, at Pin 11 Vfb is a strict no no per the reference circuit. It can cause instability and possibly
some high frequency oscillation at the regulator output. This mod is also VERY difficult to perform the soldering
on due to the location of the solder point relative to other connections is this area.. Its akin to doing Micro
Surgery with a Chainsaw with connections and traces so close together.

It should also be noted that the voltage measuring point shown for the Pin 11 Vfb mod is on the wrong side
of the resistor as well. It only make the voltage measured read low by .005 volts though. The correct way to
preform this mod is by raising the voltage at Pin 7 Vref of the CM3708 regulator. While this does require
soldering at two points on the MB its much easier to perform, if you use the correct components, because of
the location of the components you are working with.

CAUTION...Take to heart the part about making sure the the middle and outer legs of the Trim Pot you use
read full resistance (2000 omhs) initially when you connect it across R628!!! You will distroy the SouthBridge
from Over Voltage if you do not!!! A safety series resistor, to prevent gross over voltage, was not included
because even a quarter watt unit has a lead diameter that would make soldering it to the MB trace akin to
gluing a telephone pole to a quarter cleanly. Also if you can't solder don't try it!!!

This method of modding SB Vdd volts was performed by me on my board and by another circuit design
engineer in France. It works well and correctly using the specified parts, without introducing noise, plus
its relatively easy to do. If you use a 2 watt, single turn, panel mount style pot and 16 gauge wire to
connect then all bets are off!

AS ALWAYS I am not responsible if you blow the mod and kill your board!!!!!!
MarginJohn

Modification drawing as follows;




I found a good picture of the R628/R630 junction indicated the the drawing above. It was taken by nForcersHQ
member "Sysfailur". You can clearly see the trace coming off of Pin 7 of the CM3708 and running to the left and under
the upper ends of R628, the Cap and R630. You want to solder one lead from the Trim Pot to the UPPER side of any
of these SMD's. If you happen get a solder bridge between the upper ends of these SMD's its of no consequence as
they are already connected together there.




For those that think this mod is "to complicated";

My neighbors 12 year old son did the mod successfully by just looking at and reading the
this post then practicing his soldering technique before doing it. This kid has only been dabling
in electronics for a couple of months. He is a VERY smart kid but his biggest advantages are/were
steady hands, that don't shake from years of drinking coffee (and/or beer lol), eyes like a freaken
cat and an open mind. I was looking over his shoulder while he did it and the little $hit would tell
me to "shut up and let me do it myself" every time I tried to give him a tip. Turns out he didn't
need any either cause when I inspected his job through my 4x magnifiers it was PERFECT!
MarginJohn

[theonetheycall]

I already ahve links to 2 already proven ways to doing this mod. Your looks way to complicated.. And with the other mod you get 1.9v Max.

[ViperJohn]

TheOne I have those links as well. They are both Pin 11 Vfb mods. They work, its not the correct way and you
can see it on a scope, but they do work. Don't let looks decieve you. This way of modding is easier however as
I have done it both ways. If you have already modded using the Pin 11Vfb method and its working well there is
absolutely no reason to change. As for the voltage limit using this method...THERE IS NONE! Its only limited by
the B+ input to the R628/R630 voltage divider and thats about 3.32 volts.

The 1.725v starting point and 1.8v max are my suggestions for a long lived SouthBridge!

Happy Overclocking
MarginJohn

[theonetheycall]

Yeah I understand I don't take anything away form your method im going to recommend it to some people who might be interested and spread the word..



The more the merier

[ViperJohn]

I didn't take your reply that way TheOne. In reading the forums on the original Vfb method it was appearent
many people were having problem pulling it off without running into trouble of one kind or another. Of course
using a 2 watt panel mount pot and what looks to 18 gauge wire, as i saw in one picture, is asking for problems
(gees I hope that wasn't you LOL)! When people are resorting to trying to glue the wire on means its beyond the
ability of many to perform. I have (or should say had) a Nasa solder certificate and the right tools and its a b****
to do the soldering, even with 4x magnifiers, with the Pin 11Vfb method period. It requires a high degree of skill
to pull off correctly and as a result probably has a high degree of failure. One look at the CM3708 spec sheets, the
reference circuit and studying the motherboard traces showed a way that was much easier to perform as well as
being proper with respect to the way the was chip designed to be implimented. I am not taking away anything
from the originator of the Pin 11 Vfb method. He (or she) figured out the original problem of what was holding the
FSB back from higher speeds and got the ball rolling in the first place.

What voltage are you running your Vdd at and what are you getting on the FSB speed with it?

MarginJohn

[JimS]

It seems that maybe I could put a resistor of 2k ohms or less at position R541 and accomplish the same thing?

I would not have the adjustment capability, but I've soldered many SMD components and with that position not occupied, this would be very easy.

I would have to experiment to find the right resistor value though.

I really like you're idea here MarginJohn, please respond to my idea when you have time, thanks.

[ViperJohn]

It wouldn't work Jim. The upper solder pad of R541 is the B+ supply point (or one of them) to the R628/R630 voltage
divider and connects directly to one side of R628 via a motherboard trace. Not sure where the other solder pad of
R541 goes to but its not to the R628/R630/Pin 7 junction for sure. To raise the reference voltage at Pin 7 Vref you
have to either lower the resistance of R628 by paralleling it with another resistor, which is what we are doing here
or increasing the resistance of R630 which would mean removing and replacing it.
MarginJohn

[JimS]

I see what you are saying, was hoping R541 was connected properly.

R628 is 150 ohm, R630 is 130 ohm. What would I replace R630 with to give me a constant VOUT of 1.75?

I know there is a mathematical formula for this, but it would take me a week to find it. Thanks again.

If my math is right, the divider created by R628 and R630 is .86 or 1.15?
While most people are not going to try to replace R630, for me that would be much easier than adding the trimmer as I have all the soldering tools to do SMD work.

[Alcoa]

Hahahahaha, 2 posts on same subject. Had to move this here. Nice work MarginJohn! One question/request though, could you possibly post a link as to what the trimmer looks like and or where it can be purchased? Thanks for the help.

[ViperJohn]

Jim Vout from the divider would be (R630/(R628+R630))xVin or in this case, assuming your values for R628 and
R630 are correct (I looked but don't remember anymore),(130/(130+150)=.4643 so Vout=.4643*Vin. This is the
as built setup. Factory Vdd is about 1.6v so that makes the Vin=1.6/.4243 or 3.45 volts which seems high if I
remember correctly so you need to measure it to be sure (upper R541 pad to ground or lower R628 pad to ground).
ASSUMING the 3.45 Vin is correct then a R630 value of 150 ohms would give you a Vout of 1.725v and a R630 value
of 155 ohms, which is a not a standard value, would give a Vout of 1.753v.

There is the reason for using a Trim Pot. You have a problem with needing to use non-standard resistor values and
very small changes in values, just in the resistor tolerance, have large changes in Vout.

The Trimmer is a hell of a lot easier. Any decent electronics store (not Radio Shack) will have the trim pots readily
available for about $2.50.
MarginJohn