Biostar TForce P965 Vdroop?

[before]

Before the Vdroop mod...



... after the Vdroop mod

[before]

I bet the red circled controller is the VDDRII one, and the yellow could be the Vmch one.

yea; I think so

[quent]

... after the Vdroop mod

Would you please do the same graph at 1,4v ?

[before]

Ok; I'll do that

[quent]

Thanks

And please, give us the references on the chips I pointed out on this pic :

http://img245.imageshack.us/img245/2...0043799hp2.jpg

MCH mod shouldn't be harsh to find out :P

[T07N]

Does killing the vdroop really help oc'ing a lot? Isn't it a security feature for the core to have vdroop?

[quent]

Does killing the vdroop really help oc'ing a lot? Isn't it a security feature for the core to have vdroop?

Definitively no.

[Misko]

I bet the red circled controller is the VDDRII one, and the yellow could be the Vmch one.

Quent can u circle spots on this picture so I know what section of board to picture in macro mod? I need it for vcore/vmch read points?

regards Milos

[before]

Would you please do the same graph at 1,4v ?

[quent]

Hmmm these little oscillations are kinda weird...
Misko82 => It'll be difficult to find these points since I don't own the board.. You can try to find by yourself the measuring points on transistors (MOSfet), it won't be difficult

[celemine1Gig]

Looks great. Glad that it worked well.

Hmmm these little oscillations are kinda weird...

Perhaps it's due to the CPU being a dual core. The CPU needs to decide which core to use for the very operation/computation and that could cause the load not to be really 100% all the time. Look at the graph. For me it looks like depending on the current computation, the load and with it the vcore changes a bit. Only sounds logical to me. But that's only my .

[quent]

Maybe, but we would see the same thing on the graph before modding, scale is the same (0,1v<->1,9cm).
Moreover, OCCT is supposeD to load each core@100%, I don't think the CPU needS to decide anything

Edit : errors

[celemine1Gig]

Maybe, but we would see the same thing on the graph before mod, scale is the same (0,1v<->1,9cm).
Moreover, OCCT is suppose to load each core@100%, I don't think the CPU need to decide anything

You got me. Absolutely right. I didn't really have a look at the first OCCT graph.

Another theory:
As droop is there to let the manufacturers decrease decoupling capacitance, you'd most likely have to increase decoupling capacitance after doing the droop mod. The board was designed with droop in mind. To smothen out the vcore after the mod, additional capacitance could be the key.

[before]

Seems okay to me; it's just a matter of tunning...

[celemine1Gig]

Seems okay to me; it's just a matter of tunning...

[/IMG]http://img301.imageshack.us/img301/1420/2006091609h44vcoreho4.png[/IMG]

Yeah, this graph looks much better.

[quent]

You got me. Absolutely right. I didn't really have a look at the first OCCT graph.

Another theory:
As droop is there to let the manufacturers decrease decoupling capacitance, you'd most likely have to increase decoupling capacitance after doing the droop mod. The board was designed with droop in mind. To smothen out the vcore after the mod, additional capacitance could be the key.

[celemine1Gig]

BTW, just one thing that you can notice on some boards (at least I did on mine):

I have a D955XBK and I think it's the same for the well known Bad Axe for example. If you look at the row of capacitors directly next to the CPU socket, you'll see that one solder pad that was intended for another cap in the same row, was left empty. I'm pretty sure it's due cost cutting by using droop. Some manufacturers may just use lower rated caps instead of using fewer, and again some others just use completely different solder pads, because they designed the board for use with high quality or lower quality caps (two different locations for the caps, because the lower rated caps are physically taller in most cases, thus needing to be farther away from the socket area). ECS does the latter on some boards AFAIK.

[quent]

Yes, that's an totally interesting reflexion !
Caps mod could also help solving these kinda issues I think.

[wedfighter]

anyone know where's vcore measurement point??

[Misko]

anyone know where's vcore measurement point??

On the inductors (coils)

[GaS]

So what about MCH? It should be lower than it is.. 1.45V in BIOS is like 1.65 BIOS measured. Is it possible to mod this, to lower it? Because it would be great sth like 1.5V for the NB.

[Nosfer@tu]

Bump

How the hell can Vcore and Vdroop mod be the same thing ?

[celemine1Gig]

It is not the same thing. Just sharing one node. That's a big difference. VCore is FB to ground, while VDroop is Vdiff to FB.