4850 voltmod question

B

Bakku

Limp Gawd
Joined
Aug 3, 2004
Messages
317
i saw this voltmod here. for the first pencil mod, it stated in the article that it'll get around 1.4v, does it mean that no matter how much pencil lead i put between the 2 contacts, it'll get me around 1.4v tops? or would i mess up my card if i put too many strokes on it?

if someone has done this before, could you enlighten me how to do this correctly? i am pretty much a huge noob when it comes to electrical stuff.
 
The amount of strokes won't change the voltage. You're bypassing a component, so all that matters is that you make electrical contact between the two points.
 
so it's all about making contacts, not reducing resistance?

i thought the pencil trace reduces resistance and thus resulting in increased voltage, and the amount of lead would affect the final voltage output i am getting. so i am wrong about this?

if it's about making contacts, would i be able to use a circuitry metallic pen to do the mod?
 
you must measure the ohm so it goes down to the level you want.
that why its a sotmodd.
 
Zero82z said:
The amount of strokes won't change the voltage. You're bypassing a component, so all that matters is that you make electrical contact between the two points.
Click to expand...

This is not correct at all. You can absolutely change the voltage depending on how many strokes. What you're talking about is conductive ink. Pencil lead is nowhere near as conductive.

Bakku said:
i thought the pencil trace reduces resistance and thus resulting in increased voltage, and the amount of lead would affect the final voltage output i am getting. so i am wrong about this?
Click to expand...

Correct. Pick up a cheap multimeter to measure the new voltage after load. For my HD 4850, I used a #2 BIC Velocity mechanical pencil to get anywhere from 1.37 - 1.42V. I tried another standard pencil I had lying around and could only get ~1.3V.

Bakku said:
if it's about making contacts, would i be able to use a circuitry metallic pen to do the mod?
Click to expand...

Do not do this!!!! That will probably give you close to 0 resistance and you'll fry the card.
 
thx for the tips, aldamon.

by looking at this pic, do u measure voltage by placing the multimeter probes on both the measure point and ground point? but it doesnt have ground points on the pic, how do i know which ones are gounds?

overview.jpg
 
Found pic from XS Forum show where the three ground point is.

http://img.photobucket.com/albums/v238/feefea/4850vcorereadpointmm9.jpg

I heard you can also take the negative lead of your DMM and grounding it to your computer chassis, and take the positive lead of your DMM and touch it to one of the four measure points.

I have a question. I heard you need to compare the default and after shade resistance value KΩ point to get the correct 1.4v. Is this right?
 
when doing the volt pencil mod, i would only stick with penciling the GPU voltage. Dont pencil the memory voltage, its too tricky and you will kill you card by penciling too much. I did that to my 4850, i ended up pencil modding the core and that worked fine, then when i penciled the mem a little it started screwing up... freezing.... shutting monitor feedback... and then i restarted and saw artifacts and lines all over my screen. Listen guys, its all fun and games overclocking the 4850 and many people are successful as I thought i was. I find it really fun to match the performance of an expensive card. If you have not yet bought the 4850, put the extra 80-90 bucks into a 4870.


i have one on order now and i am returtning my 4850
 
lol_wut said:
There are 3 grounds on the power connector.
Click to expand...

I usually use a notch in my case for ground.

2wiced said:
when doing the volt pencil mod, i would only stick with penciling the GPU voltage. Dont pencil the memory voltage, its too tricky and you will kill you card by penciling too much.
Click to expand...

I agree with this. It seems like it's so easy to fry the memory. The memory on the 4850 will go to 1025 - 1100 at stock voltage. That's plenty fast.
 
Another thing to keep in mind is to cool the VRMs very very very well. I had found out the hard way. If any heatsink detaches a little from the VRMs and they overheat. Your videocard will lose signal to the monitor. This does not mean stabiliy issue with the GPU core or the Memory, but infact that the VRMs are overheating because the cooling is not enough. I wish i had known that, prior to ordering the 4870 and frying my 4850. Please do not mod your memory, its way too tricky, but also buy some good cooling for the card. Some quality Ram sinks. Once you invest the difference, it gets it closer to the 4870 price.

If i had known the check my heatsinks on the VRMs my card may have still been alive. looks like im out 100 dollars, but atleast ill be able to see the difference between the two cards for myself.
 
VRMs are not cooled well by the stock cooler. The same piece of metal that is touching the molten hot core and RAM is touching the VRMs and it's all insulated with POS thermal pads. Open air is better than that setup any day. My 4850 has no cooling on the VRMs and runs fine at 840/1100.
 
really? wow, maybe i did have a shitty card lol. Im not feeling that bad anymore. Anyways im returning my 4850 today, and have a 4870 on the way.
 
Yeah it was probably just your card.

However, the 4 square blocks in the VRM area do get extremely hot. I have a heatsink on them and when I touched them while running ATITool, they were scorching.
 
lol_wut said:
Yeah it was probably just your card.

However, the 4 square blocks in the VRM area do get extremely hot. I have a heatsink on them and when I touched them while running ATITool, they were scorching.
Click to expand...

Just because something gets hot to the touch, doesn't mean it's in any danger of dying or won't function at that temperature. Your finger is a lot more sensitive than VRMs.
 
aldamon said:
Just because something gets hot to the touch, doesn't mean it's in any danger of dying or won't function at that temperature. Your finger is a lot more sensitive than VRMs.
Click to expand...
They may not be in danger of immediate failure, but it doesn't hurt to put heatsinks on them. When you spend $150-$200 on a video card, you might as well take good care of it just in case one of the components is manufactured with poorer quality. If you have old parts lying around, you can make your own VRM heatsinks for free. That's what I did.

I'm just saying that even though your components most likely won't fail due to heat, there's nothing to lose from putting extra cooling on them just in case.
 
ok, i borrowed a digital multimeter from a friend and did the pencil mod, however, the vgpu voltage didn't increased as expected. i used a 2B pencil and tried my hardest to shade the resister, and it didnt work. i wonder if i did something wrong along the way. or is there some kind of coating on the pcb to prevent the pencil residue making good contacts with the resister??
 
Try a mechanical pencil. I had no luck with the first pencil I tried (no change), marginal luck with another pencil I tried (1.3V) and then it worked perfectly with the mechanical pencil I tried (1.4V+). It had thinner lead than a conventional pencil so it was easier to apply and the lead was probably a different consistency.
 
I also did mine with a lead pencil. A standard #2 lead pencil. Once again, dont kill your ram, or your card. I hope your card is not defective like mine was
 
Get a find pice of sandpaper and brush the top of the resistor. This makes it easier to pencilmod as the lead sticks better.

The stock resistance is 1.85kohm. Bring this down to 1.1-1.2kohm for a voltage of ~1.3v. You also have to remember to max out the bios voltage at 1.200v.
 
why do i have to max out the bios voltage at 1.200v? do i need to have a modded bios for the pencil mod to work?
 
Bakku said:
why do i have to max out the bios voltage at 1.200v? do i need to have a modded bios for the pencil mod to work?
Click to expand...

Because if you take the voltage to 1.2V in the BIOS you won't need to use the pencil as much to get to 1.3+.
 
Forgive me if this seems stupid, but I couldn't tell what parts to connect given the picture from the OP's link:

vddc_pencil1.jpg


Would someone please just take paintbrush and draw a yellow line or something to connect the dots?
 
Take a pencil and shade across the entire resistor. The goal is to "connect" the two metal ends.
 
I tried 3th voltmod method for increasing Vgpu up to 1.4 but I didn't find pencil in our lab, so I try to short between point with small zero resistance, when I finished soldering, the card did not shows any thing in the screen and it's Fan speed riched to highest value.
after this, I tried to remove the contact between points . then card works well but when I install the Driver in the windows, it's can't show any graphical image like windows desktop image. the Vgpu after removing zero resistance increased automatically by 0.036 volt.
also I try it under crossfire configuration in slave mode, but it didn't accept slave crossfire mode to. note that my soldering was very accurate and nothing in the board damaged in this work.
Please let me know your idea about this problem or probable solution to solve it.
what is the best solution for me in order to increase Vgpu in 4850 ?
 
ARH, you may have damaged your card. I believe the 3rd voltmod method's objective is to reduce resistance between the two solder point but not to connect them.
 
Pencil mods decrease resistance - NOT ELIMINATE IT! In other words, the amount of voltage depends on the amount of conductor (pencil lead) you lay down. Be careful, and use a voltmeter.
 
You must log in or register to reply here.
Back
Top