View Full Version : GF4 4400 volt mod
Professor12
02-02-2003, 06:05 PM
Well i decided to try the vmem volt mod for my GF4 a week ago. Everything went smoothly and has been working great since. The problem is that the volt mod did nothing. I measured the voltage on the card before and after only to find that the voltage didn't move at all. No amount of turning on the pot would change the voltage. So i decided to redo it thinking i might have screwed up soldering. Well 2 1/2 hours later i finally was able to get the damn thing back on again and booted it up. Again it worked but no change in vmem. I have since taken it off but i am wondering what is the problem? I am using a 1k ohm pot from Radio Shack set at 1K and I am attaching it to pin 13 on SC1175CSW. I then attached the ground to the back of a capacitor. This is on an MSI Geforce 4 4400. Any insight into this would be appreciated.
Professor12
02-04-2003, 03:13 PM
Anyone? Perhaps i wasn't grounding it right or are these cards really picky?
tweaky
02-05-2003, 03:30 PM
Isn't it supposed to be attached to pins 18 and 20? I don't remember anything being said about pin 13, unless the 4400 differs from the 4600 mem mod?
Professor12
02-05-2003, 07:41 PM
arrg i don't know the naming scheme but it is the 3rd pin from the left at the top of the chip looking at it from left to right. The ground is supposed to be the top one on the left or so it looks over at www.maximumoc.com. I am not even going to try this again until i get a better soldering iron anyways. The one i am using now is sub radioshack quality; the grip is all plastic held together by one conveniently placed screw that has burned my knuckles more times than i care to mention..
Professor12
02-08-2003, 05:03 PM
Alright well i went out and got myself a nice weller soldering iron and redid the mod. It STILL won't work and i am left all the more clueless. I am doing the mod exactly as shown on maximumoc.com (their geforce 4 4600 mod) except i soldered the ground to one of the solder pads. Does anyone have any ideas on how to make this work? Anyone even done this mod on a MSI Geforce 4 4400? I just broke 15k on 3dmark and now if i could only get more than 625 on the mem i think i could start working for 16k :) Besides more mem would also mean more eye candy in BF 1942 :D
X-bit labs has a guide for modding visiontek's ti4400(reference card) http://www.xbitlabs.com/video/visiontek-gf4ti4400/
And besides, CS1175CSW controls gpu-voltage, so you shouldnt see any change in memory voltage ;)
tweaky
02-09-2003, 04:07 AM
Yup thats the guide I used, except of course the Ti4600 one for me. :)
Professor12
02-09-2003, 02:19 PM
Well xbit does describe it as the gpu voltage but i have heard others describe it as the vmem voltage and its also described that way over at maximumoc.
I just found out the same professor. I was thinking of modding my visiontek ti4400 with that guide, but i dont know anymore because im not sure whether the article is excactly correct or not. My visiontek ti4400 runs 24/7 @ 300/660 and can benchmark @ 310/730 with default cooling and no mods. This baby could really run after Vmem and Vgpu mods and changing p4-retail cooler to gpu and adding some ramsinks.
Professor12
02-11-2003, 03:41 PM
Yeah well something happened when i did the mod (with my new weller soldering iron :)). For the first time the mod seemed to have worked or at least it must have to have burned out my card :(. I put in my card and started up the pc, immediatly checking the mem voltage... no change. A few sec after starting it though everything got weird and the card died. Upon further examination i found that the memory voltage had dropped to .8 v. ARRRGG well i guess thats what you get... it seems odd that the reference designs of 4400 and 4600 would be EXACTLY the same except for switching what the voltage regulators control? If the mod was successful i should have seen about 3.0 to 3.05 volts on the memory but alas no. Anyone have any insight on this?
I'm sad to hear that your card broke up :/
I'm definately going to change better cooling for my card and let some professional solder the Vmods.
Just tested 3dmark2003, resulting in 2077 with 24/7 clocks, being the highest results with gf4 ti4400 atm. When i get home, im going to benchmark with higher clocks and hope for an increase of 100-300points. The result isnt in orb yet, since i wasnt able to upload it yesterday, hopefully i have more luck today.
Hognert
02-12-2003, 10:53 PM
Originally posted by Professor12
Yeah well something happened when i did the mod (with my new weller soldering iron :)). For the first time the mod seemed to have worked or at least it must have to have burned out my card :(. I put in my card and started up the pc, immediatly checking the mem voltage... no change. A few sec after starting it though everything got weird and the card died. Upon further examination i found that the memory voltage had dropped to .8 v. ARRRGG well i guess thats what you get... it seems odd that the reference designs of 4400 and 4600 would be EXACTLY the same except for switching what the voltage regulators control? If the mod was successful i should have seen about 3.0 to 3.05 volts on the memory but alas no. Anyone have any insight on this?
Its the same thing with the creative ti4400. I know msi and creative normally uses the same design so I guess that explaines that ;). Anyways I once got the 0.8V thingy too.. Only that the 0.8V is the core voltage.. Check out the small 8 legged chips situated around the caps. They are actually transistors and costs like 1$ a piece. One of them burnt out and 0.8V is exacty what I got. Changed it and the card runs like a charm. Worth checking out.
Professor12
02-13-2003, 04:40 PM
Lol well its already been rma'd. I should have it back by the end of next week but thanks for the heads up if it happens again in the future.
Nomad
02-18-2003, 10:23 AM
Just in case you are still considering doing future vmods on a Ti4400, the 1102 controls vgpu -=pins 11 & 14(ground)=-, and 1175 controls vmem -=pins 18 & 20(ground)=-
I highly recommend using grabbers versus soldering straight to the chip itself. This makes it easier to rma if you damage the card.
Professor12
02-19-2003, 04:40 PM
Thx for the info... it just confirms that i was doing it right all along. I am still not sure what happened though I think it might have been a transistor blowing out (what hognert said). I checked later to see what the resistance was set at and it was only at 900 ohms which means that the memory itself couldn't have blown. And yeah i will definatly be using grabbers to test from now on (i had some that would work just never thought to look :( ). Ah well i should have a card back by next week and then i will see.
Where's your moral? :P You break the card youself because of the mods and then just send it back like it had broken up itself.
If you break it, you pay it.
Nomad
02-22-2003, 11:48 PM
Big words from someone with 4 posts.
Companies learn alot about their own products from people pushing them to the extremes. He pushed it a little to hard, so what....? The warranty doesnt say anything about being void if you o/c it.
Duh, number of posts doesnt count.
No warranty allows overclocking of any product, no matter it being a cpu or a gfx-card. And a volt-mod, sure you can rma it too :P
http://notlikethis.hole.fi/
Nomad
02-26-2003, 05:08 PM
My my my...do you work for computer company and trying to get more info on how we push our systems to the max?
I agree, number of posts doesnt count as far as knowledge of computer hardware goes....you could really know your stuff....I'm just saying that coming here and lecturing people about rma'ing their stuff isnt cool.
Professor12
02-27-2003, 05:45 PM
lol feel free i am not personally offended but nevertheless I do have my own objections to rma'ing it but i have no funds right now to replace it and no replacement at all. Hehe manufacturer's warranty statements ammuse me sometimes... if you look, amd states that not using the stock heatsink voids the warranty on the cpu at their discretion :)
Meangene
03-01-2003, 05:48 PM
"The graphics card is equipped with Samsung memory chips with 2.8V internal circuits and input/output buffers voltage (VDD/VDDQ). This voltage is formed by SC1102CS"
"There is a special SC1175CSW chip on the card responsible for the graphics chip voltage."
I think They are the opposite on the ti 4600. I almost did the mod wrong on my 4400, because I previously had done the 4600 mod and was not paying attention.
Anyways the mod did nothing for my ram, except give me artifacts.