View Full Version : 8800gtx Ovp
harleybro
08-04-2007, 04:21 PM
Anyone know where the OVP kicks in on a 8800GTX? Also any links to the OVP mod?
zanzabar
08-04-2007, 04:27 PM
whats ovp???, is it over volt protection
harleybro
08-04-2007, 04:32 PM
whats ovp???, is it over volt protection
yup
zanzabar
08-05-2007, 12:43 AM
wouldent that just involve looping your own little regulator into your volt mod when u did it, or are u looking for one that would go onto the gpu or memory circuits for the total voltage and not just what u add,
i guess that looping in a fuse of some sort were u would do the pencil mod might work aswell
NightRaven
08-05-2007, 01:24 AM
nono he means that the gtx has overvoltage protection. meaning above a certain vgpu, it just cuts off to prevent damage to the gpu. he is asking for the mod to override it. n fuse is for current :)
@harley.. i'm not too sure abt the gtx but for gts and ultra they come in around 1.45V for the samples i tested. however, one thing is that, for my ultra above 1.4v gives me artifacts when i use water. so thats one thing weird. too much voltage would really damage the core if u dun dice or ln2.
zanzabar
08-05-2007, 03:05 AM
that makes way more sense thx for the explanation, looking from an ocing standpoint that makes alot more sence
harleybro
09-13-2007, 07:35 AM
Bumping to see if anyone has updates on OVP mods. :D
Harley did you find ovp mod? I would also like to know :)
harleybro
09-20-2007, 11:35 AM
Nope I have nothing yet. :(
Nope I have nothing yet. :(
Pity...
But thanks all the same :)
7499richard
09-20-2007, 05:07 PM
I would like this as well, and to the guy saying that its around 1.45V....WRONG!!!
Mine will run 1.6V easy with water, and artifacts mean your running it to fast dummy, slow the clocks down. That thread made no sense, he didnt mention anything about his clock speeds or anything, and I hope, for his sake, that he was overclocking the card if he has volt modded it!
Nosfer@tu
10-02-2007, 02:49 PM
I read in some thead about the capasitor mod being the way to get past 1.6 Vcore.
Just cant find it again....
harleybro
10-02-2007, 04:04 PM
I read in some thead about the capasitor mod being the way to get past 1.6 Vcore.
Just cant find it again....
Here (http://www.xtremesystems.org/forums/showpost.php?p=2086556&postcount=5) I am thinking it only creates less voltage flux but could be wrong. :confused: I still would like to see a flat out OVP mod. ;)
Nosfer@tu
10-03-2007, 02:00 AM
Here (http://www.xtremesystems.org/forums/showpost.php?p=2086556&postcount=5) I am thinking it only creates less voltage flux but could be wrong. :confused: I still would like to see a flat out OVP mod. ;)
I think there is no need for at OVp mod, because there is no OVP at 1.6 Vgpu.
it is propperly just voltages start spiking so bad that the core crashes.
but Im just speculating
wittekakker
10-03-2007, 12:34 PM
I'm curently researching for a way to mod this card, hope it pays off. SF3D is sending me some coils as we speak, I killed mine by accident. The resistance of the coil could have influence on the OCP sensing circuit.
harleybro
10-03-2007, 12:36 PM
I think there is no need for at OVp mod, because there is no OVP at 1.6 Vgpu.
it is propperly just voltages start spiking so bad that the core crashes.
but Im just speculating
I don't believe it's the core crashing when the vgpu drops to 0 ;)
wittekakker
10-03-2007, 12:56 PM
Indeed, when core becomes unstable it will most likely crash the windows environment, PWM IC will keep on producing a stable volt line.
The GPU core voltage is being cut away once the GPU is being load too much, that makes us believe that the PWM IC OCP function is being triggered which resets the PWM controller.
Nosfer@tu
10-04-2007, 12:45 AM
Thx for explaining it to me HarleyBro :D
But in reference to that wittekakker couldent there be some logic to, adding capasitors. the OVP migth be trikkered by a small spike ?
I have seen some pics of Hipro5´s gfx, not sure if it was GTX or Ultra but it had a TON of coils horisontaly to the rigth side of the core...
wittekakker
10-04-2007, 10:24 AM
I'll drop all my info online once I finished research :)
harleybro
10-04-2007, 02:34 PM
I'll drop all my info online once I finished research :)
Thanks I for one will be eagerly awaiting it!
nelsmar
10-19-2007, 07:03 PM
what voltage are you tryng to get to? i have started up @ 1.72V on the GPU on my GTX.
cowie
10-21-2007, 04:09 AM
me too have got to 1.71 on my gts640 78c IDLE on stock h/s,card does a good redition of the sun after 1.50
7499richard
10-21-2007, 07:17 AM
yes we can run that high too but as soon as tou try and bench or anything the card we shut it self down, and in our cases not do to heat, we are wcing or more. Cowie that is very irresponsible with such a nice card
cowie
10-21-2007, 09:51 AM
yes you are right .but i had to see if i could just break the blackscreen at 1.48 i had.
i would not dream of running any father the a minute or two.
beleive me i'm happy at 1.48, more than e-nuff for 790 on core at 20-30c temps.
nelsmar
10-21-2007, 10:56 AM
wait what what load voltage does OVP kick in.... i have done a 1.65 idle and it did a 1.55 load before. i was tryng to push mid 700's but just couldnt get over that 729mhz mark. so i pushed it back down to 1.55idle voltage. i was having a nasty vdrop before. seems to be only 1/2 what iw as before now though.... when deos your ovp kick in guys?
Nosfer@tu
10-22-2007, 04:44 PM
As fare as I understod there is somthing watching the Vgpu and under loade it give our sparks becaus eof the card not being able to deliver that much power. And when that happens the 2. OVP kicks in :(
wittekakker
10-27-2007, 08:39 AM
I've been thinking about a way to adjust the over current trip point. As I don't have the PX3540 datasheet, it still comes down to guessing. I really don't have the time to test this stuff, I'm searching for people who can tell me if the modification could actually work. With the 8800GTS 320MB slightly becoming a thing of the past, I don't want this modification to get under dust.
What I got, have a look at the PX3540 datasheet:
http://home.scarlet.be/wittekakker/PX3540.jpg
The current is being logged for every PWM phase through the PX3540 controller, notice the ISEN+ and ISEN- pins. I've came across another PWM controller with the same kind of current sensing technique, though the PWM controller itself was only dual phase and thus less complex. Here is a look at the same current sensing technique but used on the NX2415 controller:
http://home.scarlet.be/wittekakker/simplefied.jpg
Some kind of snubber (RC) network is placed parallel over the power inductors, the voltage across the small sensing capacitor (1uF) will respond on par with the current flowing through the inductor. Atleast that is the theorie behind this way of current sensing. Current trip point is set by a voltage devider at pin 8, the output of the RC sensing filter is compared to the voltage at pin 8 and will occasionally trip the over current protection. In order to change the current sensing trip point we could quickly change the voltage at pin 8 (OCP) by adding a resistor to ground, but for the PX3540 we don't have the datasheet so we are left behind guessing what pin it could be. So, my idea would not be to change the voltage at the OCP pin, but rather alter the voltage that is created over the sensing capacitors. I believe that soldering a simple resistor over each sensing capacitor might influence the OCP trip point.
I searching for the sensing circuit on my GTS, here is what I got:
http://home.scarlet.be/wittekakker/caps.jpg
Anyone who knows if this could work?
Please do not try this yourself as this modifications is not tested yet!
Yesterday I had the chanche to log the voltage over the pointed out caps. Whenever the 8800GTS goes into hiccup, I did not measure anything weird over the capacitors, no unusual voltage spikes. I also logged the V12_SEN pin, no unusual behavior there.
yes you are right .but i had to see if i could just break the blackscreen at 1.48 i had.
i would not dream of running any father the a minute or two.
beleive me i'm happy at 1.48, more than e-nuff for 790 on core at 20-30c temps.
I did more tests with Hipro5's max-voltage mod, with 1.6V on the GPU I could clock it until 750MHz on aircooling before the GPU began to artifact and triggered a blue screen. I did not touch the DRAM clock at this point. Now, with the memory clocked at 1000MHz, which is still a safe value for my video DRAM, the videocard crashed with the GPU clocked at only 730MHz. No blue screen this time, the GPU volt line measured 0,18V, it is a failure of the PWM circuit, or rather a protection system on my opinion.
Will try more mods hopefully next week.