8800GT voltage mods

[gugle]

^Weird, cause pin#7 is ground. Getting a reading from pin#7 likely just requires some force with the probe as there's solder flux on the legs.
Anyways, both vMEM mods I linked in post #1204 work on your card too.

sorry, wrong measure

---correct---

Chip AP47067N or APW7067N

pin#4 + gnd = 0.649 Kohm
pin#7 + gnd = 0
pin#4 + pin#7 = 0.649 Kohm


mod reverse is also the topic # 1024?
pin#4 + pin#13 ? VR 50 Kohm
could be 30 kohm ??

[largon]

Yep, both vMEM and reverse vMEM work on your card too. But I wouldn't be too hopeful for any OC'ing gains as your chips aren't Qimonda...

Yes, 30kΩ works great for reverse mod.

[gugle]

Yep, both vMEM and reverse vMEM work on your card too. But I wouldn't be too hopeful for any OC'ing gains as your chips aren't Qimonda...

Yes, 30kΩ works great for reverse mod.

thanks Largon

I have to connect with the maximum vr?

I want to make the change to lower the temperature
since it is only 738mhz same

[CryptiK]

Thank you for that! I am definitely interested in capping my GS's now.. think I have a old ASUS mobo laying around here somewheres.. gonna check for some suitable caps.

*Thanks again Cryptik

No problem. Just be aware that the results will vary as the card designs are different. The 8800GT has minimalistic PWM circuitry, whereas the GS may have better designed circuitry. Cap mods are not known for giving big gains, but in my case it just happened to help a little.

Also with the caps, make sure they are high quality. Low ESR and high ripple caps are optimal, and make sure the capacitance ratings are not too high, as adding excessive capacitance is detrimental to the behaviour of the circuit.

[CL3P20]

^^ Understood.. my focus with capping the GS's, would be to minimize the fluctuation under load, and overshoot coming back to idle.. I am right at the verge of OCP, loading at ~1.418v ..I can run 06' with core speed @ 1028mhz.. but have to drop it to 1018mhz for 05', 03' and 01'..'er else, black screen..re-boot..etc.

If I can lower my GPUv by .02v after mod..it will be worth it. I would really love to see 1100mhz core speed

*As for caps.. I was thinking 6.3v, 1000uf - would these be a suitable choice, you think? ..I ask, because I see you used 1.5k uf ...I will not even pretend to know how the farad rating would affect, what I am trying to accomplish.

[gugle]

Yep, both vMEM and reverse vMEM work on your card too. But I wouldn't be too hopeful for any OC'ing gains as your chips aren't Qimonda...

Yes, 30kΩ works great for reverse mod.

mod completed
vMem 2.0v

Thanks

[CryptiK]

To choose the appropriate rated cap you need to measure the voltage at the position in the circuit you want to add the caps. As the input side of the card is 12v, using 16v 1500uf rated caps should be about perfect, if you leave the standard caps in place. On the output side, where the voltage is whatever the vcore (vgpu) is set to (between 1v and 2v) 6.3v 1000uf electrolytic caps will be fine. You need to choose a cap with a voltage rating above the standard voltage it is subjected to, or else it will be overloaded, but it can't be too much higher or the cap wont be in its operating range.

If you use electrolytic caps like I did, larger uf ratings are required (eg 1000 - 2200uf would be ideal) and if you use solid caps, 870 - 1000uf should be sufficient.

When you add caps to existing caps, you need to add the cap in parallel, which means the -ve side of the cap goes to -ve of component, and the positive side of the cap goes to the +ve side of the component.

In the pic below, I indicated where I measured 12v, and added the 16v cap and where I measured vcore (vgpu) and added the 6.3v cap. I have only shown one phase but the other phase is identical.

[CL3P20]

If you use electrolytic caps like I did, larger uf ratings are required (eg 1000 - 2200uf would be ideal) and if you use solid caps, 870 - 1000uf should be sufficient.

When you add caps to existing caps, you need to add the cap in parallel, which means the -ve side of the cap goes to -ve of component, and the positive side of the cap goes to the +ve side of the component.

Thank you for both of those.. very helpful info.

As for the input/output capping.. you found capping the input side to be more effective for buffering GPUv..correct? [using the 16v, 1.5k uf caps]

[CryptiK]

Well there was no change in idle to load voltage changes from adding either of the pairs of caps. Rather it's the 'invisible spikes' or ripple, that occurs too fast to see with a multimeter, that I was aiming to reduce. To see ripple, you need to hook up an oscilloscope, and study the waveform under the varying conditions. I have never seen a spike over max load voltage or a dip below idle voltage, just normal fluctuations in between due to load, and this has remained the case after the cap mod too.

The GPU & memory are very sensitive to ripple. The purpose of the output caps is to buffer, or 'smooth out' the voltage/current being fed to the GPU/ram etc. This can lead to increased stability, if the filtering was insufficient to begin with.

I reasoned that as there are already 2 solids caps per phase on the input side, and no larger output caps, adding output caps probably would give the best results. After I removed the input caps there was no decrease in performance, so in my case it was the output caps that made the slight difference. It just depends on the individual card and circuit design.

[CL3P20]

LoL.. I have a scope..but its for RF.. not sure how I would 'hook it up' to the DC though.. anyhow.. I have located some 6v, 1000uf caps, on a 'parts mobo'.. the nice black/gold ones..cant remember the brand name right now.. I think they will do nicely though.

Again, I appreciate the input, and explanation.

[Bigred1]

Well i went ahead and finally tried this vmod on my EVGA 8800GT SSC. Well the ground was easy enough but the vsense i screwed up on tremendously. The SMDs(i think there smds) highlighted in Red got to hot from my soldering iron and fell off. (this is not my card, just same pcb). Basically Im screwed right. My computer now just turns on, Shuts off, and then reboots. I just want to make sure that I messed up and theres no way to fix this. Thanks

[largon]

You could try soldering them back? Just get a smaller head for you soldering iron before trying...

Can you measure the resistances of the resistors (black SMDs)? That white SMD is a capacitor.
Polarity doesn't matter with either types of parts.

[Bigred1]

You could try soldering them back? Just get a smaller head for you soldering iron before trying...

Can you measure the resistances of the resistors (black SMDs)? That white SMD is a capacitor.
Polarity doesn't matter with either types of parts.

I would try but I have no idea were they went in the process. I went out and bought a 9800GTX+ today to replace, but if its fixable i wouldnt mind fixing it or sending it to someone who could fix it.

[Vatos_locos]

go to a local store that sells resistors and these staff show them the card and tell them that you want at least the two black smd's they are all the same in the row so you need to buy 2 that they have the same value written on them as the two next to them on the card
as for the capacitor i thing that it will work without it but if you find one put solder it back on place

[largon]

Those detached resistors are not of same resistance value. And there's nothing written on them so one can't know what to buy without the original values.

[Vatos_locos]

are you sure ? in the photo i see they have 3 numbers on them

[CryptiK]

Here's a pic of my card with the Sanyo WG 1000uf 6.3v caps added.

[manieq]

Is 2.12v mem 24/7 save on my galaxy 88gt? Def is 2.0v.

[CryptiK]

I don't think overvolting anything is 'safe', it just comes down to if your mem responds to it and you are comfortable running it.

Most quimonda mem ran better undervolted, and the samsung mem likes more voltage, depends on your card. I would try and cool the mem controller/PWM as it already get ridiculously hot at 2.0v.

[HitandRun]

Is 2.12v mem 24/7 save on my galaxy 88gt? Def is 2.0v.

I have the Samsung mem and run 2.2V for 2400Mhz for like 8 months now. Never had a problem!

[CL3P20]

I need some help locating the mod point for this non-reference XFX 8800GT, please.

I have ground continuity between FB pin#6, and the 'purple' and 'blue' marked resistors...
measured resistance of labeled points as follows:

FB-> 'red' = 4.97M ohms
FB-> 'purple' = 1.009M ohms
FB-> 'blue' = 33.1M ohms



..am I correct in thinking that 'red' *20 for my VR and mod point?

[westk]

Help guys, my XFX 88GT has a different layout and only I need do to it the vgpu vmod, here some pics:









>Here the 500ohms VR



Thx in advance!

[noroger]

Gigabyte 8800GT reference card 600/900/1500 stock
800/999/2000 after

I need 1.5V (load) to keep it stable at 800MHz.
1.336V in windows, 1.4V playing games, 1.5V using ATiTools artifact scanner. Why is there such a huge difference?

[gugle]

which maximum voltage that can use the G92 to test and which for the day-day?
and what the G86?

and what the maximum temperatures for G92 and G86?

[noroger]

not sure, mines a G92 and takes the 1.5V like a man, it depends on your cooling