Cracking the Duron/Athlon voltage code

[dimmreaper]

I made some observations from this article at Anandtech. http://www.anandtech.com/showdoc.html?i=1345&p=4
[img = "http://www.tir.com/~theevans/dimmreaper/images/code.jpg"]

Your will notice from the diagram above (ripped it off them modified it) that I have listed the voltage that is added for each bridge. You will also notice that their are five resistors and five bridges, coincidence? I don't think so. I also think I know what the other dot may be used for. I think that it is prodded after the L7s have been cut in the factor, in order to test voltage (the true core voltage, as in "1.75V" or what ever). When I get my Duron I will test the resistance of the resistors to see if they correspond with calculated voltage. I might be on to something, not too sure. Anyway the values of each bridge are an added piece to the puzzle. My hope is that someday we will learn to "hack" the voltage on the CPU rather than on the MoBo (to voltages greater than 1.85). We will see what I come up with. In the mean time, if you have any theories on anything about this, let us all know it. Keep it real, and clock on all . . . . . .

[AZZKICKER]

hey Jeff

well it sounds good

ill have to help you digg alittle

=)

[Big Mike]

From the AMD Socket A technical brief:

The VID[4:0] signals are outputs to the motherboard that
indicate the required VCC_CORE voltage for the processor. The
VCC_CORE ID (VID) is sent to the motherboard VCC_CORE
regulator. The processor VID[4:0] outputs are open drain. See
“ Voltage Identification (VID[4:0])” on page 24 for the AC and
DC characteristics for VID[4:0].

My understanding from this is that the L bridges for the voltage are switches, IE the chip just tells the motherboard on or off for the ammounts of voltage it wants. I don't think modifying the chip will gain any voltage, i think the motherboard voltage regulator has to be fooled into a higher output.
-Mike

[dimmreaper]

Big Mike (Dec 25, 2000 07:06 p.m.):
From the AMD Socket A technical brief:

The VID[4:0] signals are outputs to the motherboard that
indicate the required VCC_CORE voltage for the processor. The
VCC_CORE ID (VID) is sent to the motherboard VCC_CORE
regulator. The processor VID[4:0] outputs are open drain. See
“ Voltage Identification (VID[4:0])” on page 24 for the AC and
DC characteristics for VID[4:0].

My understanding from this is that the L bridges for the voltage are switches, IE the chip just tells the motherboard on or off for the ammounts of voltage it wants. I don't think modifying the chip will gain any voltage, i think the motherboard voltage regulator has to be fooled into a higher output.
-MikeThis may very well be the case mike. But to me that sounds like a waste of 10 pins on the bottum of the CPU packedge. But again, you may be right. Thanx for you input. I will have to go take a gander at the technical brief too. To bad I'm not the most technical guy. But I should know enough to not have to drag out a tech dictionary :)

[AZZKICKER]

jeff you tell me where to look and ill help you out bro

im a electronic eng. so i understand this ****

just dont know where to find the T Bird spec sheets and stuff
if i had those i would be able to help out bro

[dimmreaper]

check this one out AZZ http://www.amd.com/products/cpg/athlon/techdocs/pdf/23792.pdf

[dimmreaper]

looky what I found . . . .

2.3 Push-Pull (PP) Drivers.

The Socket A AMD Athlon Processor Model 4 supports Push-Pull (PP) drivers. The system logic configures the AMD Athlon Processor Model 4 with the configuration parameter called SysPushPull (1=PP). The impedance of the PP drivers is set to match the impedance of the motherboard by two external resistors connected to the ZN and ZP pins.

Does this mean that the CPU "samples" the voltage from the MoBo then sets it self to regulate the Vcore? Any input from an elite electronics guy would be helpful.

[dimmreaper]

it appears that a whooping 18 of Socket462s pins are attributed to Vcore. Talk about complex. It does however leave me to believe that it is hackable.

[Big Mike]

Well what i gathered from perusing the data sheets <It'd be nice if there was one that explained the basic motherboard interface a little more simply...> is that the VCORE pins are the actual voltage pins for the CPU, the high power demands of the Athlon were part of why it uses such a high pin count so that it could have alot of voltage pins to take in power. From what I gathered the VID pins which controll voltage are just connections to open <ground> and are simply on off switches...I had a thought on how to test the theory though, try connecting diagonally, that would, if the chip uses resistance to determine the voltage value of a bridge, allow you to run more than one pin at the higher setting, ie use .05 connected to the .025 and see if you get .1 volt extra instead of .75 extra or something to that effect...kinda hard to explain...i guess it'd be like this
ok, the text version didnt work for anything, heres the picture version...

[dimmreaper]

I too tought of something similar. But I don't know if I will try it on my duron or not. I was thinking of connect the .2V bridges to the .1V as well as the .2V in hopes of gaining .1Vs of Vcore. I may try it. I'm not really sure what will happen. Ijust hope I don't F it up. I will be getting my duron next month when KT133A boards come on the scene.

Here is what I had in mind
||\||.
Notice how the second bridge (.2V) is patched in to the bottum of the 3rd bridge (.1V)

[Big Mike]

yeah breakin the chip was sorta my concern or id be trying it right now, lol...however if either one of our theorys on how it works is right it should be fine...

[dimmreaper]

Here is my Idea, hopefully it will get me 1.9 - 1.95V

I just hope I don't see sparkes. If I do it'll be the most expensive key chain I ever bought.

[Big Mike]

Yeah that'd be a real let down...lol, too bad we cant find someone with "corporate sponsorship" to try it. If you go ahead with it let us know.

[dimmreaper]

I can only see 4 possable out comes of trying this.

1)no change in voltage
2)voltage changes, but it doesn't exceed 1.85V
3)voltage changes in the predicted manner
4)wildcard, weird voltages/perminant CPU damage

[Big Mike]

I've seen worse odds, lol....maybe I'll give it a shot when I get home...actually no ill probably be beating on my car, but if i get a chance ill try it.

[surlyjoe]

with all those pins for v-core ID ,,it seems like you could wire wrap em like I do the celly 2's? ..hmm

[AZZKICKER]

HEY GARRET WHERE YOU BEEN?

WELCOME BACK !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[dimmreaper]

cullen got his Duron we will know soon enough.

[Deviant]

Just measured the resistance of thoses resistors next to the L7 bridges.
Well they are all 0.7 ohms, and quite a large CSA for surface mount resistors. This tells me they might be current sense resistors, rather than voltage setting resistors.

[Cullen]

No no, I dont have the duron yet, its on its way from the place online. I forgot where i ordered it from...
So just sit tight for little bit jeff..

also-What exactly am i going to be doing to this poor duron? I imagine i will be wiring pins and what not. are you guys gonna tell me which pins to wire or am i on my own here?

:)

[dimmreaper]

your going to try something like this.
[img = "http://forums.overclockers.ws/UltraBoard/Public/Uploads/Post-2-456-Idier.jpg"]
Since the 1.45V, .1V and .025V come connected to equal 1.6V on the Duron. I suggest connecting the top .1V dot to the bottum .05V Get it?

[AZZKICKER]

to measure the resister correctly it must be taken out of circuit

love yours truly ,
Mr. Electronics Eng.

[Deviant]

It was, and still is outa circuit, it happens to be broken, however I doubt it's the resistors, cause a golden orb got the better of it.

[AZZKICKER]

you have to unsloder the resistors off the chip, then you check what the resistance is
not the chip out of circuit but the resistor itself

DOH

[Deviant]

I'm with you now, parallel resistances, DOH.
Seeing as it stuffed anyway, I will desolder and remeasure.

[AZZKICKER]

you have to pull resistors no matter how there wire in circuit
on a resister that has the wires on it you can sipmlly unsolder one leg and its "out of circuit"

but on surface mount resistors you have to remove them all the way so you dont damage anything

[AZZKICKER]

but i think your catching on

[Deviant]

On to the surface mount resistors. What hard little suckers to remove. Anyway bad news for your idea. They are capacitors. When measuring their resistance, my multimeter just charges up, until it reads greater than 20Megaohms. I havent measured their capacitance, but if you think it may help I will. All 5 are >20Mohm.

[dimmreaper]

any news on all of this ??? Cullen ?

[BlackOut]

Why don't you all just short curcuit the L1 pins and ajust the Voltage on you mainboard???

[dimmreaper]

BlackOut (Jan 04, 2001 04:51 p.m.):
Why don't you all just short curcuit the L1 pins and ajust the Voltage on you mainboard???It's obvious that your a little new here so I'll surpress the urge to open a can on you. ;)
The L1 bridges are for multiplier, not voltage. You don't need to short anything to adjust voltage, the boards that support Vcore adjustment will just do it. Welcome to our forum, we hope you can learn something from us, just as we hope we can learn something from you :O

[Deviant]

Lucky you got Jeff on a good day!!