8800s OVP Mod ......

[largon]

hipro5,
The GPU is watercooled and peaking @ 46ºC. Can't be the temps.
And it trips literally instantly when 3D is started, no time for temps to go overboard.
Dunno what's going on with card...
I guess the next step is to rebuild the vGPU circuitry... Got the chokes and caps ready...

[wittekakker]

George, I have soldered a 100k on my 8800 Ultra.
As i trim, it starts from high resistance and goes down.
Once i reach ~7.57k it suddenly jumps down to ~3.362k.

Is the objective to raise the resistance across that resistor? If so, is 7.6k to ground OK?

The mod explains to start at 100k, lowering resistance would do the trick I guess.

Only the vGPU mod (which has been around for ages already) is from Shamino, OVP is all Hipro's handwriting.

I'm sorry, let's say they both rock, I appriciate what they have done for this community

Maybe it's shuts down from over temperature protection or something....

Small 3D Mark 2001 runs, temps under 70°C, nat that bad considering it is aircooled and has 1,6V on it's head

Are you sure you have soldered to the same resistor?......If you have soldered an 100K trimmer, even fully opened, it shouldn't show you more than 5.12K.....

I got around 4,8k after taking 50k or 100k VR to ground, I've even lowered it to 4,2k but no better results. What I did notice that, once I changed from 4,8k to 4,2k, sudenly the output voltage dropped with 0,1Volts, even more. Not that bad, just a bit weird so tee such things happening

hipro5,
The GPU is watercooled and peaking @ 46ºC. Can't be the temps.
And it trips literally instantly when 3D is started, no time for temps to go overboard.
Dunno what's going on with card...
I guess the next step is to rebuild the vGPU circuitry... Got the chokes and caps ready...

Well, I haven't spend much time testing this new mod, only 30min, though I was working on a way to change the over current protection trip point, maybe that helps us too. Most of my free time is not spend on overclocking though, I don't know when I ever get my results online, I have to do a lot of research since I'm not an engineer in electronics

Hipro, can you please explain me/us how you got this mod working? The datasheet learns us totally nothing. It's not like a feedback loop it seems, no resistor devider, or maybe the second component is located somewhere on the other side of the board? Anyway, the resistor is connected to ground and to the PWM controller, that's all I can say for now. Hipro, would you be so kind to explain us how you got this working, I'm very interested in this stuff, I would possible learn a lot from this

[T_M]

Are you sure you have soldered to the same resistor?......If you have soldered an 100K trimmer, even fully opened, it shouldn't show you more than 5.12K.....

Measuring the resistance across the resistor prior to the mod (or to ground) i get 3.xxk ohms. Adding the 100K with it at max resistance i get 9x.xxk Ohms. Then by reducing the VR it drops down, until i hit 7.xxk then it suddenly trips over to 3.xxk.

[dinos22]

temp protection should kick in around 120C

but try lowering your GFX card clocks and bumping voltage gratually largon to see where the limits are

[hipro5]

Measuring the resistance across the resistor prior to the mod (or to ground) i get 3.xxk ohms. Adding the 100K with it at max resistance i get 9x.xxk Ohms. Then by reducing the VR it drops down, until i hit 7.xxk then it suddenly trips over to 3.xxk.

by default - as the VGA is - it's resistor IS 5.2K.....How do you measure by default 3.xxK.......?.....


Some updates:

A friend of mine did the OVP Mod on his 8800GTX VGA....Though he got a bit more overclockability out of it (he gained some more MHz out of the Core), the "powering down" still exists.....
I asked him to measure the Vdroop of it......From 1.61V it went down to 1.54V during bench....
SO I think that this is the cause to work properly the OVP Mod.....One resistor is the "start up" voltage and the other resistors "tracks" the "spikes" during runing.....So if a "spike" of Core Voltage is high, the VGA powers down.....
During Vdroop (running a 3D), the VGPU NOT only goes down BUT produces HIGH "spikes" as well during a "relaxed scene" of the 3D.....
That's the reason that it worked perfect for me......
I only have "freezings" when my VGA can't go higher.......NO power down.....During a freezing at 1.75V (idle), I STILL measure 1.72V (load).....NO power down......Maybe coz of the "extra" components I have added on it for less Vdroop....
I'll check it out.....

[T_M]

In a few hours time ill post up pics of the mod i performed on my 8800 Ultra, along with pics of measuring the resistance both with and without the mod.
Any requests?

[giorgioprimo]

I only have "freezings" when my VGA can't go higher.......NO power down.....During a freezing at 1.75V (idle), I STILL measure 1.72V (load).....NO power down......Maybe coz of the "extra" components I have added on it for less Vdroop....
I'll check it out.....[/B]

did you add some capcitors as you usaully do ?

come on ! let we know your secrets

[hipro5]

did you add some capcitors as you usaully do ?

come on ! let we know your secrets

Photo is above...... ^

[T_M]

Hipro, could it possibly be the fact that im performing the mod after the vcore mod is done? Is there any interaction?
Anyway, about to head home and will have pics up in a few hours.

[TiN_]

T_M

you can do ovp mod anytime, It will noT interact in bad way Just will give you more room for raising vcore, if you had issues with card powerdown NOT due temperatures..

[T_M]

Thanks TiN but i kinda knew that.
WHat i didnt know was why my card was behaving totally differently just on the resistances than what Hipro suggests.

Measure my card again, this time with mod removed and every single time i measure it i get a different reading. Tried cleaning the resistor up and still the same. Either 2.1M, 3.2K, and sometime flashes of 5.3K.
Have removed the mod for now, let some more guinee pigs confirm it on an Ultra. Mine's XFX btw.

[hipro5]

Thanks TiN but i kinda knew that.
WHat i didnt know was why my card was behaving totally differently just on the resistances than what Hipro suggests.

Measure my card again, this time with mod removed and every single time i measure it i get a different reading. Tried cleaning the resistor up and still the same. Either 2.1M, 3.2K, and sometime flashes of 5.3K.
Have removed the mod for now, let some more guinee pigs confirm it on an Ultra. Mine's XFX btw.

Mine's ALSO XFX......Brand doesn't make ANY difference.....You buy only .....brand and the exact same product....

Please check you battery in your multimeter.....

[TiN_]

T_M

All 8800 cards are made on two plants, and retailers like ASUS,XFX etc just put their lable\cooling on already manufactured cards.

Agree with hipro5, try to change your multimeter bat or try another multimeter.

[T_M]

I had expected those responses and probably should have told you guys already.
Im using a Fluke 114 and the battery is only a couple months old.
The probes have been cleaned with isopropyl alcohol as has the resistor.
It reads totally reliably on other resistors (e.g. the one right next to the OVP one)

Please can someone else read their resistance

[TiN_]

T_M

couple of months is pretty much time

There is no miracle, and resistance cannot jump in such big difference on it's own. Maybe you have bad cable connection between DMM and probes, check everything. Try to reverse probes, maybe your multimeter charges some capacitor on curcuitry (in measuring resistance mode DMM puts some voltage (1-2V usually) to probes, and reads difference).

[chispy]

Thanks Hipro and Shamino, I was waiting for this Mod for a long time , thank you alot for sharing your knowledge with the community.

chispy.

[largon]

hipro5,
What are the specs of those electrolyte caps you used to beef-up the vGPU filters?
Wouldn't ceramic caps have been better?

And did you remove the original vGPU inductors or are the coils "piggy-backed"?

[hipro5]

@ chispy: you're welcome...

hipro5,
What are the specs of those electrolyte caps you used to beef-up the vGPU filters?
Wouldn't ceramic caps have been better?

And did you remove the original vGPU inductors or are the coils "piggy-backed"?

According to THIS post of mine Digital VRM NEEDS electrolytic caps as well....
I have added some 2200uF/6.3V plus some chokes in PARRALEL with the existing ones so to:
a. avoid the big Vroop
b. Lower the resistance of each coil
c. filter better the output VGPU

Ceramic caps will do bullsh.. as for the "spikes" of the VGPU and filtering during load/semi-load/idle....

[T_M]

So, has anyone else tried it yet?

[7499richard]

i would try it...but it doesnt make much sense to me....and it seems as though the only one with success is OP.

I think the OVP on this card is high...because I have set mine to about 1.75V and run half way through 3dm06 and then it shuts down. The card never brok 48degrees C so it was notheat related. I think it must have been a large voltage spike or something that set it off..

Personally I think clean DC voltage is far more important for the gpu if you want stability. So beefing up the supply is important to me. That is what I want to do...so if the OP...or anybody else for that matter, has any guides or pics of how to do this pls post with part numbers or values for the upgraded components.

[cowie]

thanks *(again) for mod
will try gts mod with pencil tonight,i should easy get that res. down with pencil.
plus i need some ice cus this is no mod for air.

[cowie]

i penciled down from 5,22 to 4,88.on a a3 gts.
it lowered my core volts to 1.16 1.14 load?so i cleaned it off and restarted back to 1.51 1.48 load.
did not try to bring up volts, is this what needs to be done cus that would "bottom" out my vr no?
i will try for more findings asap

[giorgioprimo]

wich value for the chokes ?

Photo is above...... ^

[hipro5]

wich value for the chokes ?

Link doesn't work....Get some .68uH at 25A+ each (at 25*C).....

[giorgioprimo]

Thanks

Link doesn't work....Get some .68uH at 25A+ each (at 25*C).....