8800GT voltage mods

[n91htmare]

Can someone explain what OCP is and what the OCP mod does to help stability?

[Jimbo Mahoney]

OCP = over current protection.

The mod raises the limit at which OCP kicks in, meaning you can run higher voltages and higher clock speeds without the card shutting down.

[labs23]

this is very useful! and without that bus narrowed down to 256 bit for G92GT, it could have been easy to vmod a gt with 1G of memory way beyond the reach of Ultra!

[S1mon-]

I definitely need to volt mod my 8800GT. Just need to find out the best way to do it lol.

[scottc19]

How high do think its relativly safe to push core voltage as long as you can keep the core close to 50c, becuase I'm not woried about the core blowing up, im worried about breaking a mosfet or something? As far as cooling goes, i have an mcw60 on core and swiftechs heatsink set on the chips, with artic silver 5 + superglue. Right now I dont have a fan on them, but I suppose I could go find one and put it on there... Does anyone run 1.5 -1.55v?

[S1mon-]

I know I might sound a little ridiculous but how can I check my actual vGPU using a multimeter... I have done some physics in college but it's definitely scaring me hehe. Am I only suppose to touch one vGPU spot with a multimeter and it will give me the voltage? Do I need to do that when the card is running? Is there any risks that I short-circuit my card if my hand is a little bit shaky?

Thanks

[Timbosan]

Set dmm to 20VDC
black probe to ground
red probe to vgpu measure point (while card is running)

dont bridge the measure point to another point with the red probe or you may get sparks, card shutdown, damage or all of the above.

if you cant get your hand to stop shaking, try to calm down, and clapping, hard, once or twice may help reduce the shaking.

but youre only measuing something so just chill out and measure it

[largon]

The red probe can touch multiple vGPU measuring points at the same time - but nothing else.

[***Deimos***]

I know I might sound a little ridiculous but how can I check my actual vGPU using a multimeter... I have done some physics in college but it's definitely scaring me hehe. Am I only suppose to touch one vGPU spot with a multimeter and it will give me the voltage? Do I need to do that when the card is running? Is there any risks that I short-circuit my card if my hand is a little bit shaky?

Thanks

Not ridiculous at all... we all started out at the same starting line.
Scaring you... it should - your physics college hasn't gone completely to waste if you know enough that you can easily turn all those long hours of work into a paperweight.

Now back to buisness. As with anything you just need a bit of practice. Start off with some battery devices (ie DC). Next, pop open a mouse or some USB device and while computer is on, practice using multimeter set to voltmeter. USB devices are cheap to replace, and use low power (<5V, <3Watt). Now, a mouse flat on the desk it easy to "disect", but a video card inside a case is tricky.

Get some extra 4 pin "molex" (black, red, black, yellow) power extension cables, cut in middle and solder each of 4 wires to some measuring points. Glue or tape to the case or the card (vertically point up is best). So, you should have something like this :

1      2      3      4
________________
|    |     |     |     |
|    |     |     |     |
|___|___|___|___|
  |     |     |     |     
  |     |     |     |
  |     |     |     |
vGPU  nu  nu    ground

where the vGPU will be attached to measuring point, the ground attached to ground, and the nu are not used. Now you can stick in the probes into holes 1 and 4 (doesnt matter whether red or black probe), and get real time measurement of vGPU without using hands. It doesnt matter which cables or holes you use, but its a bit more dangerous using holes next to each other because the probes can lean against each other and short citcuit. Highly recommended you tape over all exposed wires that you soldered.

[largon]

The mod-guide pic shows some easy vGPU measuring points.
For example, the row of violet dots just some pixels below the caption "G92 voltage mods" is perfectly safe to poke with a DMM probe. No need to worry about shorting things as you can touch multiple individual dot at the same time - they're connected to each other anyways.

So all you need to do is stick the black DMM probe in a black (ground) slot of any a molex 4pin plug and poke one or more of the violet dots with the red probe.

[hausner]

Anyone find some vmods for Galaxy 8800GT 3phase - short PCB?
Thx!

[Renegade5399]

Doesn't that have the factory volt mod done on it? The little jumper on the top of the card?

[hausner]

Doesn't that have the factory volt mod done on it? The little jumper on the top of the card?

yeah but i need more Volts

[alexio]

I'll do the vRAM, vGPU and OCP mods next weekend with some silverglue. I'll let you guys know how it turns out
Maybe I'll even do some tests with aluminium loaded glue (I have very fine Al for painting).

[zbomb5610]

i had the volt mod done on my 8800gt running at around 1.35 volts under load w/o any problems, i even did some thorough testing but wasn't getting much stable above 780 core and 1950 shader, so i thought the ocp mod might help out, at first it seemed that the ocp mod was helping out w/ stability at higher clocks, but then the temp was rising a little higher than i wanted it to, around 70, started to throw a few errors, and then the comp shut down. The comp would not power back up again, gave a loud whining noise, everything powered on fine w/o the 8800gt power pin in but nothing on the monitor, so i popped in my 6600 and it worked fine, so i had a dead 8800gt after the ocp mod at 1.4 volts load. just thought i would let you guys know. I got the 8800gt replaced, but I doubt i would run over 1.3 volts now and I probably wouldn't do the ocp mod. Maybe I just had some bad luck, who knows, but the soldering was done perfectly. used 1.2k resistors for the ocp mod also.

[natedog420]

im runnin an asus 8800gt with vgpu and ocp mod but they are removed from the card. i run whats in my sig for 24/7 and have been running that for weeks. i ran as high as 1.48 loaded around 900 core for a lil while with no problems. just heard a little high pitch noise comin from the card at that voltage i didnt like that at all but then its not so xtreme i would say.

[jason4207]

i had the volt mod done on my 8800gt running at around 1.35 volts under load w/o any problems, i even did some thorough testing but wasn't getting much stable above 780 core and 1950 shader, so i thought the ocp mod might help out, at first it seemed that the ocp mod was helping out w/ stability at higher clocks, but then the temp was rising a little higher than i wanted it to, around 70, started to throw a few errors, and then the comp shut down. The comp would not power back up again, gave a loud whining noise, everything powered on fine w/o the 8800gt power pin in but nothing on the monitor, so i popped in my 6600 and it worked fine, so i had a dead 8800gt after the ocp mod at 1.4 volts load. just thought i would let you guys know. I got the 8800gt replaced, but I doubt i would run over 1.3 volts now and I probably wouldn't do the ocp mod. Maybe I just had some bad luck, who knows, but the soldering was done perfectly. used 1.2k resistors for the ocp mod also.

I killed my 8800GT over the weekend as well. I finally got it back from a month long RMA due to the holidays, and proceeded to bury it after only a few days! I did all the mods w/ excellent soldering work, and used 3.92k resistors for OCP.

I was running it w/ the HR03GT and a nice high CFM Panaflo fan. I put HS's on everything I could. Got it up to 864/2000/1026 at 1.36 idle. Didn't feel comfortable going any higher on the voltage on air. Temps were getting to 52*C w/ ATI Tool. I decided to run the FUR stability test b/c I know it can get the cards hotter than ATI Tool. Walked-away for a second, and when I came back the PC was off, and pressing the power button yielded no result. A flip of the PSU switch allowed the PC to flash on for a 1/2 second , but still no boot. I put my 8800GTS-512 back in and it works fine now.

I disassembled all mods and card still doesn't work. 2 possible culprits.

(1) I see that one of the wires to the vGPU mod had some of it's insulation melted away (didn't see this until I removed the mod). This could be from it laying across a hot chip/component. It may have shorted out, but I don't think this is the culprit.

(2) I cranked down the bolts on the cooler pretty hard. I could see a definite bend in the card PCB. I backed off a turn on each screw, but it was still bent a little bit. I think I may have mounted the cooler unevenly (my 1st mount on an exposed die) and it appears the corner of the die is chipped off a bit. I think the combination of the improper mount and the higher temps proved to be too much for the poor silicon die!

My 1st gfx card volt mod started out great, but ended in an unfortunate disaster!

I might post some pics later...I forgot to bring my camera's flash card to work.

Thinking it was my mount, and not some bad soldering work (it looked darn good if I must say) I proceeded to volt mod my 8800GTS-512. (I know....I'm stubborn, but I have a much better 'feel' for how to mount the cooler now) I've got it all running now w/ the HR03GT on it and the vGPU mod, and vMem buck mod. It was late when I finished all that, though, so no results yet, but the card is idling happily at 39*C w/ my previous OC of 771/1944/1062. For now the VRs are tuned to max resistance. I'll start ramping it up tonight if the wife isn't feeling too needy. I'll post results in the appropriate thread.

[scottc19]

Hmm, well I have been having better luck so far. Im at 1.5v load temps have never got above 45C yet. I have a fan running to cool the mosfets which are also heavily heatsinked... they get really hot without a fan. I also have a sink on the mosfet on the back of the card. Anyway I have been running this 24/7 for serveral days now and have achieved stable 900core, 2265 shader, 1000 memory. I hope mine doesnt crap out too? I did ocp mod with two 1.2k resistors measured them before hand and they were both 1.183k. My card seems straight with no bends and no chips. Sorry to hear about your guys cards, maybe ill soon be joining you guys lol

[zbomb5610]

I killed my 8800GT over the weekend as well. I finally got it back from a month long RMA due to the holidays, and proceeded to bury it after only a few days! I did all the mods w/ excellent soldering work, and used 3.92k resistors for OCP.

I was running it w/ the HR03GT and a nice high CFM Panaflo fan. I put HS's on everything I could. Got it up to 864/2000/1026 at 1.36 idle. Didn't feel comfortable going any higher on the voltage on air. Temps were getting to 52*C w/ ATI Tool. I decided to run the FUR stability test b/c I know it can get the cards hotter than ATI Tool. Walked-away for a second, and when I came back the PC was off, and pressing the power button yielded no result. A flip of the PSU switch allowed the PC to flash on for a 1/2 second , but still no boot. I put my 8800GTS-512 back in and it works fine now.

I disassembled all mods and card still doesn't work. 2 possible culprits.

(1) I see that one of the wires to the vGPU mod had some of it's insulation melted away (didn't see this until I removed the mod). This could be from it laying across a hot chip/component. It may have shorted out, but I don't think this is the culprit.

(2) I cranked down the bolts on the cooler pretty hard. I could see a definite bend in the card PCB. I backed off a turn on each screw, but it was still bent a little bit. I think I may have mounted the cooler unevenly (my 1st mount on an exposed die) and it appears the corner of the die is chipped off a bit. I think the combination of the improper mount and the higher temps proved to be too much for the poor silicon die!

My 1st gfx card volt mod started out great, but ended in an unfortunate disaster!

I might post some pics later...I forgot to bring my camera's flash card to work.

Thinking it was my mount, and not some bad soldering work (it looked darn good if I must say so) I proceeded to volt mod my 8800GTS-512. (I know....I'm stubborn, but I have a much better 'feel' for how to mount the cooler now) I've got it all running now w/ the HR03GT on it and the vGPU mod, and vMem buck mod. It was late when I finished all that, though, so no results yet, but the card is idling happily at 39*C w/ my previous OC of 771/1944/1062. For now the VRs are tuned to max resistance. I'll start ramping it up tonight if the wife isn't feeling too needy. I'll post results in the appropriate thread.

that sucks man, mine was doing the 1/2 second quick flash on as well and then nothing. funny thing is I was running ati tool and the fur benchmark at the same time, seems like that fur filled donut killed both of our cards.

[jason4207]

that sucks man, mine was doing the 1/2 second quick flash on as well and then nothing. funny thing is I was running ati tool and the fur benchmark at the same time, seems like that fur filled donut killed both of our cards.

Has anyone else tried the FUR donut at high clocks? Don't do it if you haven't already, as I don't want you to kill anything, but if you have already then let us know how much higher your temps went (compared to other stress tests like ATI Tool, Crysis, etc.)?

[***Deimos***]

How high do think its relativly safe to push core voltage as long as you can keep the core close to 50c, becuase I'm not woried about the core blowing up, im worried about breaking a mosfet or something? As far as cooling goes, i have an mcw60 on core and swiftechs heatsink set on the chips, with artic silver 5 + superglue. Right now I dont have a fan on them, but I suppose I could go find one and put it on there... Does anyone run 1.5 -1.55v?

Amen

I think power regulation circuitry cooling is more important than people realize. Afterall, we're no longer talking about grandpa's old GF4 days, back when video cards didn't need external power connectors and memory and MOSFETs lay naked as the almighty had made em.

Turn on a 100W lightbulb. Wait 5min. Grab and hold it with your hands. Hot huh? Using a power adapter for a for router or battery charger... the power adapter gets pretty hot no? I think its safe to assume for higher power video cards, stress and heat on power regulation circuitry increases proportionally. Even more so with overvolting and overclocking.

Its tiny 65nm. I would stick below 1.4V, and heatsinks and a fan on MOSFETS and VREG chips is a must. I haven't heard anybody here kill a 8800 by cooling it too much, or making too small voltage increase.

[zbomb5610]

so i got a replacement for the bfg 8800gt that blew on me, but the new one is a blueish green pcb and it has a white looking layer on top of all the points on the card (mainly the voltage mod point), plus this card doesn't clock as well as the one i blew which sucks.

[f117]

hi there.
A friend of mine has an Albatron 8800GT Short PCB




And i cant find anything about vgpu on that board.

Does anyone have a clue about this board?

anybody?

[largon]

Well well, my late 8800GT died during FUR benchie too...


f117,
I can't do anything with those pics. Cellphones are no good for photographing circuits, let alone for macro.

[natedog420]

my advice would be to everybody who owns or is buying a 8800gt. do not run fur bench as i havent and i have no stablilty issues or lockups at 876/20xx/1902 at 1.3 idle and 1.38 load. running great knock on wood.