where can i get the data sheet? I've tried google, but it has been no help.
BTW i did the vram mod from vr-zone, I'm running my ram at 2.37V. I can bench easily without arifacts at 1350, NFSU2 at anything much above 1200.
Printable View
where can i get the data sheet? I've tried google, but it has been no help.
BTW i did the vram mod from vr-zone, I'm running my ram at 2.37V. I can bench easily without arifacts at 1350, NFSU2 at anything much above 1200.
#301 What could you do before the mod?
And what type of cooling do You have?
BTW how mush is standard Vcore?
Would bee nice to do the VDDt mod ha ?? ;)
LOL at his username :D
Has anyone managed to further enhance their cards with additional mods or has managed higher clocks with a few cooling/voltage tweaks?
I've managed to verify glitch free clocks of 430/956 running 24/7 via bios mod, anything higher on the GPU core results in significantly lower memory clocks and vice versa, I could only run the memory @~936MHz if GPU is @438MHz...
standard vcore is 1.3V, kinda low considering it's a 110nm chip.
what's vddt?
i cool my card with a big chunky s939 HSF.
before i did vddr i could do 1120 fully stable. my core does 465
I put an slk900 and a really low cfm fan on my 6800 GT and it dropped my temp by about 20c. I only did the bios vmod to 1.4v and actually no improvement on my overclock than before with the stock cooler, but I can now overclock my memory stable to 1250mhz. I got a crap core tho :(
how have you fix your SLK900 on the board ?
zipties as suggested to me by another member (HaLDoL)
20C drop in temps LOL!!!!! :D
You need to up the GPU core @1.5v to go higher.
I was thinking something small and lighter in weight like this: http://www.overclockers.co.uk/acatalog/Others.html
bios modded to 1.5v didn't help at all, if anything it made my overclock worse I think. Hardmod might work better, but I don't know why. Maybe if I did some cap mods, and I'm interested in adding additional mosfets if someone would tell me how to do that, I know I just gotta solder them on top of the existing ones, but I don't know what type of mosfets are needed for this.
Last GT I had, I tried the vmem mod and fried it, not sure why though, it looked fine to me, but that makes me scared to try vmem or vcore yunno.
edit: would cap mods help a GT at all? either on top of the smd caps on the back, or in parralell with the 2 caps which gain power from the molex? I have added some on the smd caps before (in preperation for the second molex mod) but it didn't help. That however was on my old GT and my load temps were reaching the low 70's so that could have been limiting me. What about the other 2 caps?
edit2: One quick question, in doing the 2nd molex mod, I'm using a disc capacitor, how can you tell which side is negative and which is positive?
i tried the bios mod. it didn't help at all. I think only some cards support the bios mod.
Hey, I got an idea, not sure if it would work tho, but I don't see why not, and it would only benefit people that draw their vdimm from the 5v line, but I was trying to mod my dfi 250gb to do that, and thats how I came up with this idea. Would it work with the added molex connector, to make it draw from the 3.3v line? I mean, all you need is 1.4v+/- .1 so thats enough, but the 3.3v line has plenty of amperage, and it would be a completely different line (and if ur memory draws from 5v, it's pretty clean too). What about this idea?
I was thinking exactly the same actually, that would be really nice.
just, bumping this thread back up. I did the additional molex mod, and it gained me nothing. I'm quite sure it's working too, I measure 1.45v from it, and my card is running at 1.4v in the bios. Before I would artifact at 410mhz, now I still artifact at 410mhz. What a rip :( it's drawing from the 5v line btw, didn't mod the dfi, and I'm going 939 soon, so I doubt I will try and mod it.
here's pics, just testing it out. The mosfets don't even get hot, but testing with a dmm I do read 1.45v :confused:
http://img85.exs.cx/img85/7599/mod1fi.jpg
http://img85.exs.cx/img85/9476/card0jo.jpg
http://img85.exs.cx/img85/6498/running4uh.jpg
update: the mods not working. Am I supposed to ground the - side of the cap or something?
yes, the - side of the cap should be grounded.
-CaT
Ok I want do do the Vddr and Vgpu on my 6800nu and maybe the cap mods. What are the yellow markings in the pic for? Ground? Whats the max safe Vddr? Where can I get some 100k pots and the caps online?
So You did the power mod but "the mosfets don't even get hot" ?!Quote:
Originally Posted by kryptobs2000
One of the mosfets - they aren't mosfets btw, but PNP power transistor should get hot and the mod should be used in conjunction with 1.4V/1.5V Bios V-mod for best result.
This mod has helped me to clock my GT up to 505/1180 !
http://service.futuremark.com/compare?2k1=8455111 http://www.xtremesystems.org/forums/...es/rockout.gif
:confused:
Edit: Going through your image and the circuit/mod ... seems correct - just like mine ! :D
Here's the deal, unconnect the mod from the board and adjust Vout to at least 1.55V-1.6V instead of 1.4V-1.45V [ use 1.4V in VC-Bios ].
Once the mod is connected to the board the voltage will drop back to 1.42-1.45V for the core. The thing is, You must overvolt the circuit in order to take load off VC onboard/own power supply.
Now PNP power transistor should start to get warm/hot.
Exactly what PNP Power transistor did You used ?
ah, that makes sense. I had it running at the same (1.45v) as my board. I'll give that a try. Thanks OnLine.
Yeah and don't be scared if the screen becomes like a big mess when You post with the mod already connected to the card.Quote:
Originally Posted by kryptobs2000
It won't kill your card, but it may confuse it ! http://www.xtremesystems.org/forums/...milies/lol.gif
I recommend You to connect the mod after You've booted in to Windows and nVidia display drivers has been fully loaded and changed Vcore from 1.1V to 1.4V.
1.1V is the boot up voltage for a GT and supplying 1.55V in parallel may confuse the card a bit !
It has given me this screen of mess a couple of times.
1.4V, adding 1.55V-1.60V in parallel wouldn't [ shouldn't ] cause You any problem though.
If someone can find a fix and make a GT to work at 1.4V before Windows/nVidia display drivers loads, then great ...
Btw, how is your DFI NF3 250GB Vdimm hardware mod going ?
I'm very interested doing such a mod myself ! :stick:
Well it's working, and no problems when booting up and only having 1.1v, at least yet anyways.
And about the dfi mod, I havn't done it, and I don't think I am going to because I'm thinking about selling the board and if I do, I don't want to mess it up. Going 939, just waiting for newegg to ship my k8ns ultra.
edit: Well it is working I'm pretty sure. I think I've gained about 5-10mhz with it. I was hoping for more but I'm only using 1.4v anyways. I'll vmod later and see how it does, hoping for at least 450mhz on the core (410mhz now).
Just did the vgpu mod, it's now set at 1.6v however I get no additional overclock, if anything, it is now worse. Whats the problem? Temps are fine, 48c idle, not sure load, havn't checked yet.
http://img82.exs.cx/img82/411/gtvmod8vw.th.jpg
edit: I havn't tested it with the missing molex mod yet, my fan connector broke off so I need to fix that. My only guess is that maybe I am not recieving enough current from the single molex, I'll try setting the additional molex mod at 1.7v and see how it works.
edit2: Forgot to followup. With the additional molex, I still get no increase in overclock at all. I find this very wierd that i gain 20mhz+ going from 1.3-1.4v but I gain nothing but occasional throttling from 1.6v :confused:
Yeah it's working alright.Quote:
Originally Posted by kryptobs2000
1.6V on the core is high, You should try 1.5V instead and set the mod to work at 1.60V [ and slowly increasing the voltage up to 1.70V if necessary ].
This mod isn't perfect for higher voltage than 1.5V ~ the load between the the additional power and the boards own power supply becomes very unbalanced at 1.5V[+].
Try to disable FW. Perhaps You'll need better cooling like water cooling for higher voltage and speeds ! http://www.xtremesystems.org/forums/...lies/crash.gif
You don't think this is keeping it cool enough lol?
http://img129.exs.cx/img129/8094/gtcooling4th.th.jpg
It's not making the best contact but it should be doing it's job fine. I can't get 1.5v core unless I take off the mod. The lowest I can go with the mod is 1.6v by flashing to the 1.3v bios. This mod raises my vgpu by .3v, turning the pot(10k) does nothing. Not sure why lardarse didn't say to use a higher rated pot. I will try adding a 22k resistor tommarrow and hopefully that'll allow me to use 1.5v.
On another note, I tried the 1.5v through bios flash, no hardmods done at the time, I gained nothing from it. Also, doing all the mods, bios flash aside, this includes cooling mod, missing molex mod, vgpu mod, I have gained absolutly nothing on the core, how dissapointing. :(. I think my card just sucks.
edit:
Just for reference here's my experience so far:
Stock, 1.3v bios I would get 385mhz before artifacts.
Flashed to 1.4v (1.43 measure from multimeter) I would get 405mhz
All mods done 1.59v actual, 415 gets artifacts, 400mhz freezes up (throttles, power problem?)
Right now with all mods done, 1.4 actual (1.1v bios + .3 hardmod) 398mhz
I'd really like to get at least 450mhz at least out of this card. It's really a shame the core just really sucks or something on this. Any way you can see to improve this molex mod to allow 1.6v+ maybe?
Is there any way of modding the bios in my GFX card to go higher than 1.4V?
Or does anyone know which points to do a pencil mod on it?
what is your card ?
is there any more simple power mod for the 6800gt???
i understand that the vmem and vgpu mod on vr-zone should work with 6800gt's (256mb), but i'm also aware the power circuitry on the gt's isn't as complete as the ultras... i've seen the whole extra molex power thing - i'm incapable of making something like that...
but if there is something i can improve my gt with, ie... solder on a few capacitors, a mosfet maybe, i'd be willing to do that...
i've skimmed through this whole thread but can't exactly find the answers to my question... so can anyone direct me to a link somewhere? cheers
if you look closely to the schematics of the extra molex mod, you will see it's not a very difficult mod ;)
ok... i'll look into it...
OK well i just did the vddr mod, far out that resistor is tiny... but yeah, got it working... at 500kohm on the vr i get 2.22v for the memory already, and it didn't improve my OC's... so i dialed it up to 2.3, same deal, up to 2.35v...and now i've realised it has DECREASED my OC... :mad:
why has this happened? ram is samsung 2ns... this is a 256mb 6800gt...
damn... now i wanna try undervolting it ehehe...
curious if anyone else has found that using a multimeter measuring at TP11 (6800NU gigabyte) stock 1.2v bios gives 1.2v, mod it to 1.4v and the meter reads 1.3v and give it 1.5v and measured is 1.4v.
wondering if anyone could take a look :) ty
http://www.xtremesystems.org/forums/...ad.php?t=57645
Have heard that nibitor won't change bios so correctly if you change to 1.4 so it will be 1.3. On my geforce 6800LE the volt is 1.41-1.42 with a 1.5v bios and heard that the 1.6v bios should be 1.47v. They really undervolt :(Quote:
Originally Posted by Lesterp
that's fine with me, i only wanted 1.3v - 1.4v is what seems to be killing GTs and NU cards.
nbitor change nothing on my board:'(
but the capacitor mod for the mem worked great, 10mhz without an increase in voltage
hello,
i searching for a easy to solder Vmem Vmod for tha ASUS V9999GT 128MB AGP videocard. Anybody know a guide or Infos how it works ? I Only can found Vgpu Vmods
Here this card:
http://www.ocer.net/article/a_upload/1103796908_4-.jpg
http://www.web-upload.de/data/media/21/2.jpg
thanks for help
Is there a point on the back of the board where I can add an extra core cap? I can't add one to the front like in the guide as the Arctic Silencer covers that whole area..
poke around with ur multimeter on cap legs on the back of the mobo and find one that reads w/e ur core voltage is (not sure what card you have so I can't say). Then just solder the cap in parralell with that one.Quote:
Originally Posted by markiemrboo
So, I could use the core voltage measure point + ground?
6800NU, sorry :)
No, well.. it depends, but to be safe just find a capacitor on the front of the board, flip the board over and measure the + leg, if it measures the same as ur core it's more than likely a cap in that circuit, so then just solder ur cap in parralell with that one.
All the caps look to be just soldered to pads on the front of the board... they don't 'go through the board', so I can't really do that.Quote:
Originally Posted by kryptobs2000
I tried holding a probe to the + of the cap and then flipping it over and probing at random points. Lots of the points beeped at me, but when I measured the voltage of some of them later, none of them seemed to measure 1.2v (core voltage)
Anyway you can get a pick of the back and front of the board? Back should do but both would help.
Just serch for Tp11 on the back of your card. There you can add a capacitor. But don't think you need a extra capacitor for only 1.2vQuote:
Originally Posted by markiemrboo
Quote:
Originally Posted by ZeroHero
There IS a tiny resistor (resistance) between TP11 and core caps. I measured value ~ 1 ohm from GW6800le (see the 'Black screen quide)
http://www.xtremesystems.org/forums/...ad.php?t=59391
You may think that 1 ohm does not matter, but the series resistance of those caps might be ~ 0.05 ohms, and extra resistance to those nodes cancels the effect of added capacitor.
So, TP11 is NOT good point to solder extra cap and you must use capacitor leads for that purpose. However, I think also that extra cap for core is not required..
Original GT bios @1.3V
http://service.futuremark.com/compare?2k3=3925517
Ak350 (Silver + nickel plated) + Delta 60mm
@452/1150Mhz
http://img192.echo.cx/img192/3609/gpu6800gt3ns.jpg
fotinhas da 6800Gt
http://img22.echo.cx/img22/7427/6800gt1l7hq.jpg
http://img229.echo.cx/img229/6301/6800gt2l9he.jpg
I added a cap for the core voltage and although I got me about a 10mhz gain on the 1.4 voltage (Got me up to where im able to go past 405 at a maximum of 415 in rthdribl) I dont think its worth it.Quote:
Originally Posted by SikaRippa
Given the awkwardness of how you have to place the cap cause of the NV5 cooler, Im more worried about it getting knocked off instead of how much extra mhz I can get out of the core. Im quite satisfied at keeping it at 390 infact.
Its the ram seems to make the major diffrence. I got that sitting at 830 right now although I can push it to 850 without artifacting still. I just wish I could get me some fetsinks.. Those lower fets on the back of the card get so hot they burn to the touch.
okay, good to hear the teoretical part of my idea. :DQuote:
Originally Posted by SikaRippa
I think the LE has very stable volt, my gw6800le at least just vary 0,01v with 1,41vcore. But if you hardmod to get really much volt perhaps.
With 1.3v I can get to 425MHz without seeing any artifacts in 3dmark03, so I use 420MHz for general gaming, 100MHz with 0.1v, on air, isn't too shabby :) I pushed it up to 440MHz, the ONLY place I saw artifacts there was on the turtle shell, white blinking patches :( Hesitant to try 1.4v though as I hear other components on the back of these NU's get very very hot running higher core volts??
As for my memory, well, that's on going. Trying to get that sorted out in SikaRippas thread. The extra cap isn't helping me on that one and for some reason even with higher memory voltage (2.8v) the nvidia driver overclock test STILL caps out at the same 835-840MHz as it did before, at 2.7v :(
Quote:
Originally Posted by Reznik Akime
I agree, one can achieve some extra MHz by adding an extra capacitor, but I think that main limiting factor in the most cases is the GPU itself: how good/fast chip you have !Quote:
Originally Posted by ZeroHero
As posted by ZeroHero the supply voltage varies only little 0.01 V, but the memory voltage can have VERY high ripple voltage. I mesured 0.1 V ripple with multimeter under heavy load !! Since multimeter bandwidth is very low the actual ripple can be much more higher. Why such high ripple voltage ? I have investigated it and one important factor is that core voltage capacitors (c139 and c140) are high quality componentents from Sanyo (very low ESR and high max ripple current). The memory capacitors are from 'unknown' -brand and nobody knows any specifications for them ESR= ? , max ripple current = ?
ZeroHero: could you measure the ripple voltage of memory and GPU supplies under the same (heavy load), in order to make some comparison between cards ? (I wonder you have added capacitor = filters the largest peaks ?)
I have before measured volts, my vmem is 2.71-2.72 under a 3dmark05 round with a added capacitor to the memory. Without the capcitor the volt was 2.71, but with the cap added it can go up to 2.72 under load(strange, but perhaps it's the multimeter, don't know if it's quality).Quote:
Originally Posted by SikaRippa
Vgpu 1.41 but can go up to 1,42 sometimes under a 3dmark05 round. No capacitor on gpu.
So to me either vgpu or vmem have more than 0,01v ripple.
Wat card have a vmem ripple of 0.1v?
markiemrboo: disable the clock test, if it dosen't artifact it's won't hurt your hardwere. You overclock with rivatuner or? If not try Rivatuner
I measured values once more gain to be 102 % sure :D :DQuote:
Originally Posted by ZeroHero
--
I am now measuring so called 'GW6800le version 1b' which has a stock 3.19 Vmemory voltage !!!!! (see 'Black screen guide for details), but about the same AC-values was measured with 'GW6800le version 2' (=has memory voltage of 2.72 V)
When measuring DC-voltage (!!) very stable values are obtained for both memory and core, max variation is at range of 0.02 V
BUT when measuring AC-voltage the case is totally different. Core voltage variation remains low ~ 0.02 V all the time e.g. when running 3dmark03. Memory voltage varies A LOT in respect of frame rate (=current consumption, see BS-guide again) and I got again the same maximum AC-voltage of 0.1 V !!!
--
ZeroHero: Could you verify that measurement yourself (multimeter in an AC-measurement position) ?
3.19v is crasy but good for overclockers who dosen't wanna mod. But saw that already in your BS guide.Quote:
Originally Posted by SikaRippa
Tested all in DC, can test it in Ac too. But it will be on the weekend and now i dosen't have any capacitor so perhaps it will vary.
No need to test/measure, since the problem (voltage of 3.19 V, NOT BS-oddities) is SOLVED:Quote:
Originally Posted by ZeroHero
!!! THAT CARD WAS FAULTY THEY HAVE MADE A SHORT ABOVE ONE OF THOSE CAPACITORS IN FACTORY = "GOLDEN SAMPLE" ?????? !!!!
.. I fixed it :banana: and now I have stable ~0.02 V (AC) 2.72 (DC) voltage in that node (I will post details to 'Black screen guide').
(But there IS still an AC-voltage in that node having frequencies above multimeter capabilities. I have measured them with oscilloscope from non-faulty card = I had just BS and it is *feature* for GW not a faulty)
--
For extreme overclockers this was some kind of benchmark: Has anybody before tested with a 3.19 memory voltage ?
Good, that you could fix it. Did it overclocked better with 3.19?Quote:
Originally Posted by SikaRippa
Haven't seen someone having 3.19 yet, just 3v on a LE or Nu. But one guy had 3.4v on a 6200, but that's slower memmory and on teh 9800pro/XT you could have realy high volt, someone had even 3,5v but don't think it helped so much with such high voltage.
That was a nice indeed. I wonder those memories would not stand for long time so high voltage. I can not understand HOW such card was sent to customer (=fault was not detedted) !!!Quote:
Originally Posted by ZeroHero
I did not OC too much .. little over 900, and I did not test the limits.
--
Yes, I think that those 6200 etc. cards have tatally different memories (too lazy to check now), so they have different 'absolute maximum ratings'
I can't messure the vdimm from TP1, anywhere else i can try?
You can measure Vmem (Vdimm ?) from TP1 or TP10. Please take a look this thread:Quote:
Originally Posted by Razzz
http://www.xtremesystems.org/forums/...ad.php?t=62651
..where a much more easy way to modify Vmem is presented. There are also posted some OC-results for Vmem and Vcore.
great thread, just wish it was summarized a bit...
Getting ready to mod my Leadtek 6800LE, probably go with a bios mod 1.4-1.5v and a Vddr mod + maybe the cap mod.
C
Quote:
Originally Posted by charlie
I agree nice thread, but too looong.. Maybe, Del Gro Badeu could update the first post :rolleyes:
At least I would suggest to include following figures, since the provide better way (easier etc.) to do modifications than originally proposed:
Capacitor mod for 6800le (to avoid 'Black Screen'):
http://img69.echo.cx/img69/4061/capmod35gz.th.jpg
Vmem mod for many 6800/6800le-cards:
http://img216.echo.cx/img216/4646/vmemmod16ir.th.jpg
(original thread for this, having also comments about Vcore modification)
http://www.xtremesystems.org/forums/...ad.php?t=62651
I will speak to hime about this.
He has alsow made a frensh version that is much more updated. I will make a translation coworking with him. Dident care that much but now I goot the Pci express version of the 6800 nu so Im up for modding again :D
I will report back when I hear from him :)
EDIT: Just spoke to him, he has very littel sime at the moment because of a summer job finasing his studies :)
But we will descus what to do with the thread tomorrow when he has more time.
I will use my time to clean the thread up, but would like some help. So if anybody could give me a hand and cume up with suggestions for what posts to put where.
1. An updated guide.
2. An results part.
3. An FAQ part.
4. And offtopic.
I have problems with my card :( It's a Leadtek 6800GT AGP. On 1,3V everything is ok I can o/c on AC to 415/1150 and to 440/1190 on peltier+WC. When I change voltage to 1,4V I get screen corruptions, and resets.
Capacitor mod for 6800le (to avoid 'Black Screen')
Migth need somthing like that?
never head of problems with past 1.4 Vcore.
maybe you need capasitor mod?
Maybe :) When I can find that for 6800GT ?Quote:
Originally Posted by Nosfer@tu
Trie to look throug the post. se the last 10 posts.
I wonder Leadtek 6800GT have totally different PCB (=card) and design i.e. capacitor mod does not help. In the case of BS computer freezes right a way totally not only some corruptions.Quote:
Originally Posted by luks__pl
Maybe (just maybe) adding a capacitor to Vmem or Vcore could improve OC-capability.. But your card (=chips) might be now at OC-limit.
Peltier element sounds cool cooling solution (literally!)
im not convinced that the leadtek GT has a different PCB.Quote:
Originally Posted by SikaRippa
Im not sure I understand what you are saing my freind ;)
Take look e.g. this foto (is the same card your were talking about?)Quote:
Originally Posted by Nosfer@tu
http://www.ultra-price.by/id5054.html
.. and compare it e.g. GW's 6800le card
http://img106.exs.cx/my.php?loc=img1...ge=fig36qf.jpg
Where can I find voltmods to mem and gpu for Leadtek 6800GT. I´v tried the one on VR-Zone (the Asus version for the mem volt), but it killed the card. I allsow want to have pictures where I can put the cap mods om my card. 12v 3300uf is the best cap?
I wonder the mods shown VR-Zone could be used with Leadtek 6800GT, since it has the same PCB (?)Quote:
Originally Posted by Razzz
What cap mod you mean or what you are going to gain with cap mod ? The cap mod for 'black screen of 6800le-cards' is not compatible/nor needed for cards having this PCB.
I'm looking for the same :)Quote:
Originally Posted by Razzz
Quote:
Originally Posted by SikaRippa
Yes, it looked like it have the same pcb but my card doesn´t start any more so I think it killed it. Are there any volt mods for the 6800gt/ultra exept the one on VR-zone?
What did you in detail: a pencil mod or changed resistors ? Problem with pencil mods is that it soooo unrealiable: final resistance may vary a lot from target value and it can also change when time goes. Could you measure your current resistance and voltage values (DO NOT MEASURE RESISTANCE WHILE POWER IS ON)Quote:
Originally Posted by Razzz
Ahhh you are quite rigth :DQuote:
Originally Posted by SikaRippa
A Long time since I partisipated in this Thread :p:
Just a quick question, where is the measuring point for Vgpu??
Oh and the cap mod worked like a charm! Great mod!!!
C
You have a Leadtek 6800le (?). Take a look this figure:Quote:
Originally Posted by charlie
http://img148.echo.cx/img148/3621/6800tps28bf.th.jpg
TP1=TP10 (memory voltage)
TP11 (GPU voltage, small resistor 1 ohm)
TP3, TP5 (VTT/VREF voltages)
TP9 = GND (also some other TP's)
.. this foto is taken from GW6800le, but your card should be similar.
--
Did you get 'black screen' feature with your card or was the mod only for OC'ing ?
SikaRippa,
Your mods and pic's have been exemplary.... thank you!
And the cap mod stabilized my OC'ed clocks. I'm stuck at about 900 benchable at the default 2.7v... but this was still sufficient for the new 6800LE WR in '05 with approved drivers.
C
I have the same problem (Leadtek 6800GT)Quote:
Originally Posted by luks__pl
at stock (1.3v) everything is fine, and i can do 425/1110 stable (1.35v measured). If i flash a 1.4v bios the card gets unstable and the screen corrupts under load. I replaced the crappy thermal pad on the small heatsink over the (?)vregs with thermalpaste and now the heatsink gets nice and hot under load. 1.4v works for longer now, but i still get screen corruption if i loop Gametest 4 of 3dmark03 :(
Guess the regs are just too weak :(
Is there a way to pencil mod (temporary) vmem for 6800nu AGP?
Not exactly pencil mods (AFAIK), but VERY easy anyway:Quote:
Originally Posted by ***Deimos***
http://www.xtremesystems.org/forums/...ad.php?t=62651
.. if your card is using that PCB.
Ive digged trough this whole thread trying to find a mosfet/caps/... mod for the 6800GT 256mb AGP, and I did found info, but because I'm a noob at electronics I diden't understand it.
What I'm asking now is if you guys could give me some (highres)pictures of a 6800GT on which I can see where to solder the caps/mosfets/... and what caps/mosfets/... it must be.
Ill take these pictures to a vmodder I know and he will solder the stuff for me.
Sorry to ask these "do everything for me" questions, but I must still learn a lot about electronics.
Thank you very much,
JochenP
You know that Vmem mod where you put 5-10K resistor between TP13 and TP14.. well now I have a dead 6800.
I found a 10K resistor from old pcb, unsoldered it, measured again carefully, bent the pins so that they went straight down, inserted into TP13 & TP14, and powered up. Correct expected 2.76V showed up on memory (vs 2.71V). But no beep, no boot no nothing.. fans are spining, and all the TP show correct volts for memory and gpu. Dunno what went wrong :/
Did you allready try to unsolder it again Deimos?
Also I have a dead 6800GT PCI-E. My father Vmodded it for me, and this is how it looks like: http://users.skynet.be/fb254489/IMG_0422.JPG
The Vcore regelator has an 8k resistor to ground so the card get's 1.4V (with ultra bios). With an self added 100k resistor in parallel the core wil get around 1.68V, because the total resistance lowers to around 7,4kOhm. On my picture u see also an 1M variable resistor in series with the 100k resistor. Lowering the parallel resistance will give higher Vcore, so I've added the 1M var. resistor so I am able to regulate between 1.68V and 1.4V, but I'm not able to go above 1.68V so I can not overvoltage the core by turning to quick on the variable resistor.
This should work fine in theory.
Real life went little different... When al was modded I messured the Vcore resistor. With var resistor @ 1MOhm I got +- 8kOhm, and with var resistor @ 0MOhm I got 7,5kOhm. Great, it works! But when I installed the card in my pc nothing had changed. I messured Vcore and got 1.53V. That's because I still have the 1.5V bios mod. Hardware mod is not working at all. Deinstalled and messured Vcore resistor and got 8kOhm no matter what resistance the variable resistor has.
So I guess bad contact. Luckely my card still works. So then I went to do some soldering myself. After few attempts I gave up. Aldo I was able to solder the resistors on the card, I was always having some kind of bad contact because when I messured total Vcore resistor I always got 8kOhm :s
Then I just put the mod resistors on the right spot, messured... everything good, I'm messuring 7,5kOhm... Okay, some glue on it...and...wait...
F*cking glue didnt went dry at all. After 15minutes I gave up. My hand hurts from staying in same positions that long... "Maybe I should give my daddy another shot" I thought. So I cleaned up as much glue as I was still able, and then put card back into PC without Vmod.
Booted, nice... butt wait...no screen :(
Everything stays black, monitor stays in standby, but pc is starting up and loading Windows. Messured Vcore: 0,8V
F*ck!
What can I do about it??? I'm pretty sure it's only that one resistor wich regulates Vcore.
Also this mod sucks because we had a hard time finding a good spot on the Vcore resistor to solder on... Is this normal?
I wasted a sweet 9500np that softmoded perfectly to 9700 doing something similar to vgpu. I was soldering to legs of vreg chip and slipped and got some solder under the legs... I omfg almost fell out of the kitchen chair! I just couldn't get the solder out from there. I tried holding it upside down and holding soldering iron to max on that spot... weird contortion.. Eventually got it off but look like I slipped and burned a bit of the pcb... what happened after that was week long story of card working and not working and back and forth till I ran out of bypass legroom.
ok i'm currently reading through the whole thread(on page 8 right now) since i'm getting water on this card soon and want to do the vmem mod and also the voltage mod for the gpu(1.7v's maybe). i have a couple of quick questions to direct me in the do it or not do it variety. i have posted a couple of pics of my temps and voltages as recorded in ASUS SmartDoctor for reference to my personal current temps with a tornado blowing on the stock copper heat sink(i have it right on it so there's a dead spot i need to clear by finding a way to move the fan out a little).
1. what is the max temp i should see on average for the ram.
2. i have the asus 6800gt that came on an Ultra PCB so i already have the extra molex connector and caps that came with the Ultra so do i still need to do the cap mod?
3. is it safe to use AS5 when i put individual heatsinks on the ram and every thing else that can take a heatsink?
that's it. the rest i'll keep reading and learning what i need to do for my GT to get my o/c up on both the gpu/mem. i've had it as high as 460/1170 with fastwrites off and 425(or 430 can't rememer)/1140 with fastwrites enabled. currently running bios modded 400/1100. thanks for any helpful replies. and now for the pic:
http://img359.imageshack.us/img359/1...ed1copy1mv.jpghttp://img299.imageshack.us/img299/9...ed2copy4ht.jpg
http://img396.imageshack.us/img396/8362/v9999gel1am.jpg
You have an incredible card. Consider yourself fortunate... stock, on air - wow. Is that like 460/350 = 31.4% faster over stock?Quote:
that's it. the rest i'll keep reading and learning what i need to do for my GT to get my o/c up on both the gpu/mem. i've had it as high as 460/1170 with fastwrites off and 425(or 430 can't rememer)/1140 with fastwrites enabled. currently running bios modded 400/1100
1. Max on RAM... without hotspots even 60C should be fine. Those BGA chips dont put out as much heat as the GPU.. in general if you keep the GPU cooler, you'll notice the bga's cooler too (after all, the heat just conducts throught the numerous traces)
2. Cap mod is for people who get black screen after running game for a few min... its worse in some games, and hard to notice in others... best theory now is that its caused by inadequate capacitors on 6800le and 6800nu - so, it doesnt affect you.
3. I personally dont know, although I have put globs of AS5 between NV5 and my 6800, for both GPU and the bga's. I'm wondering if perhaps I got some under them, since get internal clock test failure pretty close to stock speed... I got some very nice heavy duty fine tool cleaner
As long as you keep the stuff on the ceramic/heatsink then its fine.. try to avoid bridging contacts.
thanks for the prompt reply. looks like i'll just be doing the hard volt mods for the mem/gpu only then. just waiting till i get the water setup ordered and in my hands.
oh on the 460/1170 that isn't a regular occurance with this card. i was able to bench 2k1 and 05' with it but 03' was a no go also. sometimes a freeze then continue or straight up error out. i'm looking to get 440/1180-1200 all benches stable and cool to run 24/7.
waterblock for the gpu and sinks for everything that even looks like it can take one. even if it doesn't need it. also maybe some kind of mod to have a quite 120mm fan blowin over the back of the card to keep things even cooler. i'm not rebuilding until the AMD M2 so it's gotta play everything until then 1600x1200 with at least 4xaa 16xaf:D
Hey kryptobs2000, I saw you soldered some caps to your 6800Gt but unsoldered em again. Was this because they screwed up the card or why? And what caps did you use?
Thnx,
JochenP
Although cooling the gpu, and mem will lower overall temperature of the card, dont forget about all the voltage regulator circuitry. Specifically, there are 2-3 IC at bottom right corner of the card which get very hot. As well, on opposite side there is some transistors that get hot and are covered by heatsink for 6800ultra.. dont know bout your card though..
i'll have to check it out then. i know it has that back plate for the heat sink but is there something else that covers the back of the card that your talking about? i still plan on heat sinking everything and having a 120mm blowning on the back of the card.
can someone make it a little clearer what i have to do to do the vmem mod?
i.e. say where exactly to solder the variable resistor to, what resistance it needs to get a safe bump in voltage and what the measuring points are. my card is an inno3d 6800nu. im more than capable of modding the card, ive done motherboards and other gfx cards, i just find this guide very hard to understand what to solder where.
if required i can get a pic of my card and someone could mark where im to solder.
thanks
Hi,Quote:
Originally Posted by zipp0r
EASY Vmem mod for many 6800/6800le-cards. Is your card similar ?
http://img216.echo.cx/img216/4646/vmemmod16ir.th.jpg
(original thread for this, having also comments about Vcore modification)
http://www.xtremesystems.org/forums/...ad.php?t=62651
Yes, I am talking about all the little IC chips in SikaRippa's previous post... specifically the bottom ones which get the hotest. If you have good heatsinks on them your "ambient" temperature read through nVidia driver should decrease a lot. However, be very careful, as all of them are not offset the same level from the board. As suggestion, if you have old Athlon/P3 heatsink or f you have a nice chipset heatsink you could cut up and use that...
You are absolutely correct: the components of this card are running very hot. In fact not only the MOSFET's but also capacitors and coils.Quote:
Originally Posted by ***Deimos***
--
For those really interesting about temperatures.. I have measured skin temperatures of relevant components as follows. I installed (with glue) a Pt100-sensor (very linear/accurate electrical temperature sensor) at the end of wooden stick. I pressed extremely gently manner this sensor to the top of each component for proper time (for stabilization of temperature, ~ 10 s) and measure corresponding resistance. The measurement error should less that a few degrees, and I suppose all values are HIGHER that measured. I will later 'publish' all results in 'Black screen document', but here comes preliminay result acquired from MSI 6800le-card.
Below is a foto of measured temperatures for a idle state and second of during heavy load (Aquamark03).
http://img312.imageshack.us/img312/8...emp18ul.th.jpg
It can be clearly seen that the load has a huge effect to some components. Also the temperature of MOSFET's powering Vcore indicates (three IC's at the very bottom) they are running VERY hot i.e. some heatsinks would be nice.
Just imagine if you were a designer of these PCB's and you had the luxury to make limited edition run of specially catered to overclocking boards. Very good clean signals, negligble crosstalk or coupling capacitance, and user redundant components... Why have 1 FET heat up to 70C if you can use 3 and have them only heat up to 40C. Having a special limited edition board like that would be really sweet! :^)
But, till then, we have to make do with what me have here.
Quote:
Originally Posted by SikaRippa
it looks the same, here is my card front and back:
http://img24.imageshack.us/img24/303...copy2on.th.jpg
http://img185.imageshack.us/img185/7...copy9qc.th.jpg
can you draw on my photos, where to connect the variable resistor and where i am to solder an old molex connector so i can measure the voltage. i still dont understand your diagram :o
i think i have some variable resistors here, a 1k ohm, a 5k ohm (might be 500ohm tho cant remember ), and a 10k ohm. which one should i use? what is the default voltage? 2.7v ? what should i aim for 2.8v? and would that get me to 900mhz on the ram? its stuck at 800mhz at the moment.
thanks
I am going to regret doing this, since if you obviously can't follow SilkaRippa's excellent guide then you probably shouldn't be doing this in the first place.
http://img.photobucket.com/albums/v6...a/vddr-mod.jpg
Measure at blue test point, connect POT across TP13-TP14. Please read SilkaRippa's guide carefully (if you still haven't read it).
it just confuses me thats all, but i understand from you now, where to solder to and measure from. :)Quote:
Originally Posted by ***Deimos***
from his graph, is my default voltage, 2.758v ? and i need to set VR to 4700ohm to get 2.822v or 3300ohm to get 2.85v. which do you think would safely get me to 900mhz?
thnx
There are at least two alternatives WHY you acquired different valueQuote:
Originally Posted by zipp0r
- Measurement accuracy
- Your card HAS higher stock voltage
I wonder the absolute values of resistors that set voltage to VREF node are not too different from those I (and many others) have measured. Therefore, you could make the mod quite safely with a variable resistor having a HIGH initial value (>20 kohm) AND measuring Vmem value while adjusting resistor. A fixed resistor (e.g. 6800 ohm) in series makes this modding even more safe.
If you want to be ABSOLUTELY sure about resistance values: measure resitances between TP14-GND and TP14-TP13 and post those values here.
Remember: DO NOT MEASURE RESISTANCE WHEN POWER IS ON
I cant see any perfection on this topic- if a card has AC cooler you cant see those TP points and please where is the core regulating VR in your picture?I also cant see why things must be told as vdd+vddq- isnt there a more human way?Quote:
Originally Posted by zipp0r
Hi,Quote:
Originally Posted by soodent
Nobody is perferct ... BTW what figure you mean ?
--
Here is figure for TP's
http://img148.imageshack.us/img148/3...tps28bf.th.jpg
TP1=TP10 (memory voltage)
TP11 (GPU voltage, small resistor <1 ohm between TP11 and Vcore)
TP12 (FB for Vcore, NODE for VMOD !! resistor between TP12 and GND)
TP3, TP5 (VTT/VREF voltages)
TP9 = GND (also some other TP's)
Really i am confused about this topic.I tell what i get so far.
1.Only one regulator is needed(100k) however first post says 2.
2.In first post some yellow dots- dont know what and why.
3.This 1 regulator controls both core and mem volt but somehow everybody is talking about Vmeme or Vcore changes in their posts (not both at the same time).
4.Vvdq- unknown to me.
5.Exact value of regulator depends on spesific cards components- only logical thing yet.