CATCH22ATPLAY Got it.
Anyway looking forward for some intersting modZ from ya man.
Printable View
CATCH22ATPLAY Got it.
Anyway looking forward for some intersting modZ from ya man.
Replaced Mosfet #STS8DNF3LL for Vref. Got stable 1.40v on Vref now. Booted to desk top no problems after transplant. Ran 3dmark2001 & 2003 all the way through with no problems. No Vmods active at this point. Turned up mem from 363 to 390Mhz. Ran 3dmark2003 and half way through got green, blue and black boxes accross screen. Rebooted and desktop came back as normal.
Maybe I got too excited to see 3dmark run all the way though with out the matrix. Any suggestions??
I dunno dude. I should have left that IR7311 in. I decided to swap it with that FDS6898A and wow at no artifacts up to 440mhz then it died. I got 800mv and it's jumping around like mad. It was looking so sweet too. I'll have to see if it's still alive once i put that IR7311 back in.
MrCrusher did you swap it for the excat same part? If so maybe the reason it failed was do no another component. Also did you check what the volts were when it went wack on you? I'll be looking into it. But for now it's game time. Good thing i haven't returned that herc R9800xt and will probably be keeping it.
CATCH22ATPLAY btw, how did macci get his 471 on mem just with vdd?
Catch22,
Got 10ea #STS8DNF3LL from Digikey yesterday. Now my Vref voltage stays at 1.40v all the time.
Now that I have brought her back to life I had a problem when sometimes I would boot and get the matrix. All my voltages seem to be stable and right at stock with very little movement during 3dmark2003.
I did what you suggested and mapped out the mosfet and the ISL6522. Using the ISL6522 Specsheet I got an idea what voltages should be wHere. After several boot ups and measuring voltages I traced the problem to Pin #10 on the ISL6522. At boot up pin 10 was at 10.5v. On the Pin 10 feed is R93 which is a 10ohm resister and it was ohm'd at 1.5K. I pulled a #100 resistor from the r9700 and replaced. Now pin 10 reads at 12.3v
I have booted 20 times and no matrix and all voltages are stable. Been running 3dmark2001 in loop for 2 hours and no problems. I think the re-animation was successful.
MrCrusher Why did you make 12.3v? What's that about?
2PiZza, Nothing special about the 12.3volts. That is just a liitle less then what my PS is putting out. Pin 10 is looking for 12v at boot up to turn on the mosfet. THe resistor feeding pin 10 must have partially melted due to being over driven when I upped voltages on Vdd, Vref, Vddq so it was reading 1.5k ohm and keeping the boot voltage close to the threshold to turn on the Mosfet to power the memory.
More importantly what I think this means is that there are probably allot of cards out there that can be brought back to life with a little patience. I thought I fried the memory but it was actually the circuitry controlling the memory that fried so there is still hope for resuscitation.
Oh i can't agree with you more. I think a lot of theese dead cards only have a fried mosfet, Cap, Diode, resistor or whatever. It's great that your card is back from the dead. Nice going dude.
Now that you checked your resistors and basically every pin out of the mosfets and the isl6522 chip i decided to do the same. That one resistor you were talking about is ok on mine. It's a 10 ohm resistor. I checked everything and all ohm out fine. Since my card freaked out when i O/Ced it a bit and only put out 800mv or so i'll assume it's in reset mode because it's hitting current trip limit. I know how to fix that. But i don't really want to change that trip point. That sucker may have saved my ddr. I sure hope it did. But damn dude, I swear you never seen a cleaner 440 bench in your life. So this tells me it's Vref holding theese cards back as opposed to vddq. Vref is causing the sticks on wings and crap like that and the matrix. I was fairly sure Vref was causing the prob and i believe i've proved it. VDD will still have to be raised as needed though, like normal. Since i modded my vdd to put out more watts i'll stay at 2.725v max In fact for the 440 bench i was at 2.700 vdd. After some more experimenting i'll show you guys that mod. It's simple to do. Although maybe a bit dangerous.
MrCrusher did digikey have a mosfet with just a slightly higher amps rating and slightly lower RDSon? Made by the same manufacturer, STmicro? I just checked the one i installed and reading the data sheet i probably never should have tried it. Not a close match at all. Oh well another night i guess. Back t my game. Damn is the weather nice. I wonder if, maybe, ya.
Surf's up dudes
i have done 3 mods now on my herc 9800xt
de vcore mod de 1596 mod en the vdd mod...
de max i could get out of my xt =
545core(2.2v) & 436mem (3.2v)(hynix) cooled with ddmaze 4 20c water
anything higher then that would freeze mycard ......
That's a damn good o/c for 20C water imo.
OK my dead r9800xt is alive and kicking again. I'm using a 4804 off my r9700pro now. Seems to work the same as the original one. BTW the 4804 doesn't seem to be a Vishay nor an Infineon chip. No idea who makes it. But on bottom side it says.
LOPVC2
0942
2K05
and what looks to be a trade mark INDO with 2E02 next to it
Front side says
4804
0206
Well whatever it is it's working fine. About the same as original one before it fried. Might as well really give it a test playing my game.
Whoa is me. Still got some problems with Vref. Screen gets horizontal very fine lines when i first boot the comp. After 10 seconds or so it's fine. I'm thinking i better get the exact part that was on this card. I haven't noticed any issues after it warms up though. Plays games fine. i'll play with it this weekend and see if i can kill it again. ;)
MrCrusher, care to sell me one or 2 of those mosfets?
Catch, No problem I'll donate to you 4 #STS8DNF3LL and 4 #4804 for current and future experimentation. The best I have been able to do is 530core(2.21v) & 430mem(2.85v). This is with some dots and artifacts and completely clean at 520/420 all on non coolled h2o swiftech 22501 kit.
I just got yesterday one of these new high OC p4 3.0c over at the Fry's fountain valley. Stuck it in and I am running 15x267=4.0G!! so now I am at 8274 in3dmark2003 and 9000 is next. I going to need some serious cooling on the cpu/gpu to get to 9000.
I found this new way of stabilizing the vgpu mod today, and i'm going to do that mod...i've got only this questions...
does r1596 has 2 sides? what i mean is that if one side is to reading the ohms and the other to the solder point???
The pot that we will introduce is regulated to 20k or 40k???
Thks!
MrCrusher I got 8193 in mark03. 526/425, p4 2.8@3.66(14*260), OCZ 5:4 2-2-3-5@208, p4c800-e deluxe.
And I got 22275 in mark01 on this system.
r1596 ain't helping much. Card starts to whisle also.
er what tune does in whistle?
Ta MrCrusher and you're lucky with that 3.0c. and pretty good clocks on that card with just water.
CATCH22ATPLAY I mean the sound is like high frequency whistle. I'm talking about r1956 mod.
http://home.comcast.net/~rbmccall200...8/site1530.jpg
Ok, i figured that part of the resistance is the solder point...
Now...it's in the same place where we solder that we need to pick before on a multimeter and we should read aproximatelly 40k???
Then the pot that we will introduce in the mod should be regulated to 40k or 20k?
Thks!
VoRtAn_MaDgE you are talking about r1596? You should read about 41k. Put a 50r POT dialed at 40k. So you'll get 20k as a result. I tried to run this at 9k. It's ok.
Thks i figured it out before u answer...
I did the mod 5min before...
I will post some pics ...
Regard
Vortan
Pics...as promised:
http://www.pcdiga.com/Imagens/VoRtAn...able mod_2.jpg
http://www.pcdiga.com/Imagens/VoRtAn...able mod_1.jpg
The quality isn't the best but for a 2.1mp digi cam i can ask much more of his macro LOL
Regards
Vortan
MrCrusher ... Excellent rehab on that dead Vcard! :thumbsup:
Now, get busy with that new chip! :D
C22@Play ... quit foolin around ... no surfin till you fix your card, dude! :p:
Hehehehehehe, I gotta get some action Hobocrow.
2PiZzA, a high freq whistle? Hmm, that don't sound good at all. You're the first that's had this problem. It doesn't change the way your card receives volts etc. It just changes the current limits reset point. If you did some of the other mods that i did to the core then if it talked back to you that might be normal. Maybe you're stressing a component. I will have to assume it only whistles when you raise v core above 1.95 or so. Mosfets do switch usually from 200khz up to 1mhz. Is your 3.3v line kinda crappy? About all i can think of atm.
VoRtAn_MaDgE, nice dude. Hope the mods help.
CATCH22ATPLAY
I did just 3 mods. vcore, vmem and r1596. 3.3 is nice. I got about 1.97 on core when it starts to sing. :) Not very loud btw. I know that these chips which control voltages are pretty interesting things. They got oscillators and some other things. I'm really thinking that it's near it's limits of working.
Also no matter what resistance I put on r1596. 8kom or 22 or anything.
I was getting this sound on my 9700pro also. And I know people who were also.
Maybe your machine is so loud that you can't hear it. :)
I forgot something. It starts to whistle only under load.
I'm the one who as to say thks to u for sharing such usefull information...THKS m8!!!Quote:
VoRtAn_MaDgE, nice dude. Hope the mods help.
I'll try later to see if improves stability and a few more mhz on the core ;)
Regards
Vortan
:ROTF: C22@Play ... Like your new sig. ;)