Printable View
So, where's the memory voltage read point again? I've measured between GND (case housing) and TP10 but the voltmeter reads 0.00V (i scaled it to 20V). It's the AOpen Aeolus 6800LE @ 12x1,6vp, currently on 370/900 on nForce3 Ultra motherboard (finally got rid of stutter effects).
Edit: I even tried to measure between GND and TP1 as suggested in topic 'all 6800 vmods...', but recieved zero result.
Please help!
Hi,Quote:
Originally Posted by tiborrr
Are you sure you measured TP10 and not TP9 (=GND) ? Take a look this figure:
http://img148.imageshack.us/img148/3...tps28bf.th.jpg
--
How did you get rid stuttering ? RivaTuner-settings e.g "force constant performance level to performance 3d" or FastWrite=off or .. ?
Yes, i have doublechecked. I've measured between TP10 and case housing. Gotta check my multimeter if it still works...
Stutter effect solution - use nForce 5.11 drivers (for nForce3 chipsets) and the latest VGA drivers (82.xx).
Edit: Should I try to measure between TP9 and TP10 instead of GND(case housing) and TP10?
OK, i've succefully read the Vdimm voltage. It's set to 2.75V on my Hynix AF-28. I think i'll mod them with variable resistor up to 2.9V and see if I can reach 1GHz clock barrier.
Hi!
I'am first time on this forum. Hello everyone:)
I did v-mod on my memory like S_A_V (3.15V), and the bios core was modded by bios to 1,5V. The card now working on 480/1000 16p6v with out artefacts. Core is coolnig by water block and in load temp. is ~50*. The record of my card is 500/1020 atrefacts free and 13287 in 3D Mark 03. My card is Leadtek WinFast 400 6800le 300/700.
ORB 2003 link
Orb 2005 link
Thanks for all mod witch a found here:clap:
(Sorry for my english)
That's a nice core clock for being a LE!
Hi,Quote:
Originally Posted by Maxim
Quite impressive results, indeed :up:
However, I think having Vmem =3.15 V you are 'playing with fire' ;) Following aspects might cause problems at that voltage level:
- Vmax for memories is 2.95 V (3.6 absolute maximum rating)
- Temperature of memories, do you have standard heat sinks ?
- Temperature of MOSFET's driving Vmem
(removed double post)
Hi!
On the mempries i have that revoltec radiators REVOLTEC and FAN 80x80. All mosfet's have small radiators cooling by 50x50 and 40x40 fans.
I'am planing to give memories somethink about 3,5V or more:eek: . On my Ati9500 memory has a 3.3ns and with 4.2V they working at 760Mhz with active cooling ofcourse.
SFmE - Sorry for my english.:)
I would love to look at the pics posted in post#1 but I can't :( I get a message from firefox that the server redirects me in a way that never ends ( eg perpetual loop? ). Could you please repost the pictures?
Same goes for s_a_v's picture off his molex mod, think it sounds easy enough but need some visual proof before I do this :)
Here it comes again:Quote:
Originally Posted by MartianInvader
http://img216.imageshack.us/img216/4...mod16ir.th.jpg
(Maybe there is some problem in server, but I'll check&fix the link in the post #1)
thanx imageschack servers me better then that other server you had. Do the pics from S_A_V load where you are, if so please repost them to because I really love to see his 'easy' moley mod :D
I know it is an old post but S_A_V could you please repost your pics of the easy mod? And juding from the first post you'd need a fixed resistor in series with the vr and you seem to just have the vr? Is that because you took a chanche there or is it perfectly safe?Quote:
Originally Posted by S_A_V
The Vmem mod S_A_V made is electrically exactly the same as explained in that figure/guide I have posted. You need ONLY one resistor (either fixed or variable), putting those in series is only to avoid risk of shorting nodes with variable resistor.Quote:
Originally Posted by MartianInvader
So, the most easiest way to make this mod is get a fixed resistor having value corresponding Vmem value (see the graph in the figure) and stick that between TP's. If you want to adjust Vmem replace fixed value with Vres.
Great I will try it tomorrow, don't have much time today and am running some last unmodded benches :D
New ( not submitted results yet ) are:
06 2216
05 5430
Clocks were 2655 cpu 241 2.5-3-3-5 ram 450/950 vc :)
03 bsod on me while benching 460 core, think it's to warm ( 460 benchable only with very cold weather ) but still for all air their nice clocks.
Will edit in links when I rebooted!
Edit:
05 link http://service.futuremark.com/compare?3dm05=1856608
06 link 'There was a problem..... ':mad:
:nono: almost killed my card :nono:
Did the mod as described in post #1, with a vr as described in S_A_V's post but it didn't work. With the vr in the fanconnector, no matter what I set the vr at, I always got 3.3Xv from vref, without vr it was 2.73v.
I decided, it's cold outside so put the windows open. Got 14c ambient, thought well 3.3Xv for one run won't kill it. Started windows, disabled most crap, rebooted and set clocks,lod tweak with RT and loaded 06.
Imidiatly after switching to 3d, screen became all garbeld up, almost black but small pieces off GT1 showing flikkering ect so I pulled out the plug ( faster then powerswitch :D )
Turn power on, didn't boot. Swap vc, booted fine. Treid vc in another pc, didn't boot. Beeped like there was no vc installed. Getting desparate, thought final try, put it in the pc and it booted without problems. I think I slightly damaged them though. Will have to wait and see after playing some games if I now get bs while gaming :(
Only thing I can think off is... vr was broken ( only 3.3Xv no matter turning left or right ), hard to believe all 4 vr's I had were all broken, or I put vr wrong way inside fan connector :S That's why I asked S_A_V to repost pic's off how he did it, wanted to check everything.
Anyway, took everything off again :(
(I already posted some comments to 'BS-thread')Quote:
Originally Posted by MartianInvader
Are you 100 % sure you do not have a short/low resistance between TP's ?? A short between those nodes causes Vmem increase to 3.3 V.
What is resistance between TP's WITHOUT any added resistance (power off while measuring !!).
About Vres. There are three nodes and a maximum value (say 20 k) is allways present between two nodes. The screw adjusts the amount of resistance between third node in respect of two other nodes. So, having 'maximum adjustement' makes one res is zero while another is maximum, clear ? Therefore it is quite easy to make mistake by measuring resistance between e.g. #1 & #2 AND solder pins #2 & #3 some minutes later :rolleyes:
Think maybe the fanconnector might have a short then idk...
But without vr in the fanconnector no short, because still 2.73. Then with vr in place, it was 3.3v. Lol maybe I had wrong legs off vr connected? I used legs not below know, when vr is infront off me and the screw is to the left I use the two right legs. I didn't measure resistance, never thought of that lmao forget question I will measure res when I'm done with breakfast ;)
Measured resistance off vr's... strange thing happend. When multimeter set to 20Kohm Max value I get is 002, I set multimeter at 2000ohm and get 023. Storeclerk told me 100% 20Kohm vr but... I measured between two outer legs, Persivore told me that would always be max resistance and got 023, between inner leg and outer leg I got values between 003 and 023 depending on how much I turned the screw. I think I got the wrong vr's? Will go to the shop tomorrow and try to get the right one's, going to take multimeter with me when I go to the shop and measure there ;)
sounds like they sold you 20ohm VRs, which would act almost like a short.Quote:
Originally Posted by MartianInvader
:slapass: them for us
Yeah could have been allot wurse then putting 3.3v through my vram, wanted to use them for a mobo vcore mod lmao would have freid my cpu within an instance :upset: :upset:Quote:
Originally Posted by cirthix
Quote:
Originally Posted by MartianInvader
Extremely bad *service* from your supplier :slash:
--
This case shows, what is the gain by using a fixed resistance in series with variable resistor. Another general hint for Vmodding. It is good practice to allways measure resistance between nodes BEFORE your are going to add a resistor and resistor value of course. After mod measure resistance of nodes again, and if the value has changed A LOT (or near to 0 ohms) remove mod carefully and start again.
--
I hope your memories were not damaged. In fact I was measuring a GW 6800le -card, which has almost similar manufacturing fault as a new !! The memory voltage was 3.3 V due to a short by manufacturer :slobber: This voltage level was about some weeks and did not destroy the card (and has been working over half year after I fixed card).
Well the mem didn't like 3.3v in combination with 1001mhz but maybe I just went way over the limit. Another strange thing, now I was ataching measuring wires I thought it would be nice if I could measure vcore also and what do you know: even when flashed with 1.5v, I measures 1.1v so gues I will need to do another mod there ;) And would be surprising what my core would do with more juice if it runs 450/460 on just 1.1v !!!
The absolute maximum value for Vmem is 3.6 V (according datasheet), so I wonder 3.3 V did not cause permant damage.Quote:
Originally Posted by MartianInvader
Bios-modding programs are not 100% reliable, and it is quite easy to make mistakes. Since you have core frequency of 450/460@1.1 V, I think there is not too much reserve left - no matter how high Vcore you would have.
i did the mod too and it works very well.
now running at 3.08V at 980mhz (490mhz). i think i will go up to 3.25V and look how far i can get. 1ghz should be possible. my hynix 2.2ns ram is up to 900mhz at 2.95v in the specs so 10% overvolting with proper cooling should be no problem at all.
i use a zalman vf 700 cooler and the ram heatsinks. some little heatsinks for mosfets, too.
but i think you should not use a fixed resistor for the vmem mod. i adjusted the VR to exactly 2kohms just like in the picture but after soldering i could only get to 2.95v. with a fixed resistor i would not be able to get lower than 2kohms and be stuck at 2.95V. now thats what i call f*cked up. so adjust your VR to 2kohms, solder it and then re-adjust.