Problems with my voltmodded HD4850

[ghost_recon88]

So I had my Visiontek HD4850 voltmodded before I left for the summer. Then it was working fine. When I came back, I was gonna throw a Zalman VF1000 on it, and so I wanted to epoxy some copper RAMsinks on so the RAM would run cooler as well. So sanded the sinks down, epoxied them on, put the Zalman on, booted it up, and below are pictures of what I'm seeing on the screen :(:confused:

I'm pretty sure all the mosfets have some form of heatsink on them, so I'm really not sure what could be causing for the weird graphical glitches. It goes as far as to the Windows XP loading screen with the blue bar, then as soon as its about to get into the desktop, it just goes black. I'm sure the fix is easy (or at least I hope so). I just migrated to a laptop, so if no one here can help me get it fixed, I'm just gonna pwn it off in the classifieds.

[sHape oF gReY]

Check your VGA/DVI cable make sure is secure. Check the mod for any loose wires. If you did the mod yourself you could also try removing the mod to see if that's whats causing the problem.

If you used AS epoxy check that its not bridging any connections as it is conductive.

[ghost_recon88]

Had Jason voltmod it, and I did use use epoxy. There is no way to remove epoxy is there?

[jason4207]

I would say try to put the stock HS back on, but I'm not sure about that epoxy stuff.

The vmods are easy to remove...just pull the VR out. If that doesn't fix it then I think sHape oF gRey may be on to something.

Is there anything that can dissolve the epoxy? I had no idea that stuff was conductive. Can you see any excess epoxy that may be bridging some of the IC legs?

[secure]

If you want to break the epoxy bond, than put it in the fridge(freezer) for abit, than just twist the sinks off, although if you epoxied the whole memory sink you might pull out the chip, if the 4 corners are the only portion with epoxy your in luck with the fridge(freezer) trick.

[The Watcher]

U probably crushed part of the core...tough luck man.

[Essenar]

If he crushed the core, he wouldn't see anyting on the screen. Artifacting is a sign of heat issues usually.

[jason4207]

Can you boot into safe mode?

[Mr.Guvernment]

you said you did sandy, you sanded the ram chips?

check around all areas you worked on make sure you didn't get anything on some contacts or hit something else with your sand paper.

[The Watcher]

If he crushed the core, he wouldn't see anyting on the screen. Artifacting is a sign of heat issues usually.

Sometimes you can if the edges are damaged.

[ghost_recon88]

Can you boot into safe mode?

I'll try.

you said you did sandy, you sanded the ram chips?

check around all areas you worked on make sure you didn't get anything on some contacts or hit something else with your sand paper.

No, I sanded down the top of the RAMsinks so there would be enough clearance to get the VF1000 on.

[Essenar]

That epoxy idea was a bold move. I would NEVER epoxy anything to a video card. You just never know when you're going to need to remove it.

[sHape oF gReY]

That epoxy idea was a bold move. I would NEVER epoxy anything to a video card. You just never know when you're going to need to remove it.

Yep, learned that the hard way back in the X800 days, ended up pulling up a VR and a ram chip trying to take cooling off. Thermal tape all the way from there on out.

[Dragonprince]

The fact that you have any video at all leads me to believe you probably have something shorting out somewhere. I would vacuum the card off carefully with the brush attachment and or use a new paint brush to sweep off the entire card being careful not to be to rough. If that does not fix it you could try removing the volt mod and see if that clears things up. Not sure where you got the idea to sand the ram chips but I would not do that to fit any heat sink...you should have sanded the heatsink if it needed adjustment....

[jason4207]

The fact that you have any video at all leads me to believe you probably have something shorting out somewhere. I would vacuum the card off carefully with the brush attachment and or use a new paint brush to sweep off the entire card being careful not to be to rough. If that does not fix it you could try removing the volt mod and see if that clears things up. Not sure where you got the idea to sand the ram chips but I would not do that to fit any heat sink...you should have sanded the heatsink if it needed adjustment....

I think he sanded the sinks, not the RAM chips themselves.

[sHape oF gReY]

1. Take the cooler off and check the GPU for chips, cracks or any other obvious sign of damage.

2. If the v-mod is a pot trimmer + fan header, remove the pots see if it fixes anything, although I doubt that has anything to do with it. Also check for loose wires.

3. Go over the whole card checking for damage. It's possible (though highly unlikely) something got damaged (SMR etc. getting knocked off) during the whole process.

[ghost_recon88]

1. Take the cooler off and check the GPU for chips, cracks or any other obvious sign of damage.

2. If the v-mod is a pot trimmer + fan header, remove the pots see if it fixes anything, although I doubt that has anything to do with it. Also check for loose wires.

3. Go over the whole card checking for damage. It's possible (though highly unlikely) something got damaged (SMR etc. getting knocked off) during the whole process.

No damage at all to the actual core. Jason said if I could remove the mod by taking off both the vGPU and vMEM VR, which I already did. No damage at all anywhere on the card. I'm willing to bet it's something not being cooler properly vs. actual damage to the card or anything of that nature.

Classes start tomorrow and I've already got the desktop that its in parted out. I'm just gonna list in the classies, take a loss on my investment, and hopefully sell it to someone who has the time and patience to remove the epoxy (if that's what it takes) and get it working again.

[ghost_recon88]

Got the RAMsinks popped off fine via the freezing method, mosfets on the other hand...not so lucky :(

[sHape oF gReY]

Ouch...

It might be possible to re-solder them, hopefully.

[BossBorot]

ouch indead :eek:

[sHape oF gReY]

Thats one good reason to use thermal tape instead of epoxy. Been there done that.

[ghost_recon88]

Thats one good reason to use thermal tape instead of epoxy. Been there done that.

Problem is even the thick thermaltape doesn't hold up these solid copper RAMsinks, they're too heavy for it.

[deathman20]

I used on my 8800GT some super glue. That worked like a charm on heatsinks. Was relatively easy to remove as well just using some slight pressure to move the heatsinks up off the ram and Vregs and pop off it came. Might suggest using some of that if you don't think tape will hold it.

[Falkentyne]

Yes all it takes is one very tiny dab of super glue (or two extremely tiny ones in two opposing corners) to get the job done. you want to apply as little as possible, just enough to secure the sinks to the ram yet still have it as removeable as thermal tape. You have to be very careful though.

[Mr.Guvernment]

wouldnt appling to small dabs leave some open gap on between the heatsink and the ram piece not allowing them to make contact...

[xtkxhom3r]

wow....... hey look at the bright side....at least it wasnt a 4870 or a 4870x2 lol

[ViperJohn]

Check your VGA/DVI cable make sure is secure. Check the mod for any loose wires. If you did the mod yourself you could also try removing the mod to see if that's whats causing the problem.

If you used AS epoxy check that its not bridging any connections as it is conductive.

AS Epoxy is not conductive but it is capacitive.

Viper

[ViperJohn]

Got the RAMsinks popped off fine via the freezing method, mosfets on the other hand...not so lucky :(

Those are not mosfets. They are the Vcore series inductors. The mosfets are square chips the to right in groups of four.

Viper

[cubu]

i had that same problem with my old x800xl.... found out that i had over-tightened my waterblock, i loosened it up a little and it worked fine

[frankkesel]

had similar thing whit a 9700pro was the power suply

[El<(')>Maxi]

AS Epoxy is not conductive but it is capacitive.

Viper

How can you have one without the other? In order to store current it has to conduct it.

[ViperJohn]

How can you have one without the other? In order to store current it has to conduct it.

No...they are two distinctly different properties.

Viper

[El<(')>Maxi]

Yes but you understand my question right? How does a charge move 'into' a capacitor without any conductive action occuring? As far as I know, it's impossible.

[ViperJohn]

Yes but you understand my question right? How does a charge move 'into' a capacitor without any conductive action occuring? As far as I know, it's impossible.

A conductor conducts any voltage. A capacitor will pass an applied AC voltage or signal (down to
cutoff frequency) but not DC. Two very different properties.

If you get AS5, which is capacitive, under a memory chip it doesn't conduct the DC Vdd and Vddq
voltages to ground or to other pins like a conductor would but it will capacitive couple the data
pins, which have an AC component, together or to ground or to the Vdd and Vddq pins and it will
be INSTANT display corruption.

Viper

[El<(')>Maxi]

Yeah I know thats a no no, I'm just trying to understand the physics. I thought the main caps on a graphics card are there as part of the buck converter, storing/smoothing DC output voltage...at least thats how I understand it. DC capacitors right?

[ViperJohn]

Yeah I know thats a no no, I'm just trying to understand the physics. I thought the main caps on a graphics card are there as part of the buck converter, storing/smoothing DC output voltage...at least thats how I understand it. DC capacitors right?

I think it is more technical terms and thinking. When you apply a DC voltage to a capacitor current will flow through the circuit into the cap until the cap is fully charged. At that point current flow into the cap ceases. When the supply voltage goes lower than the charge in the cap (basically making it an AC component) the cap will discharge into the circuit in an attemp to maintain the DC voltage at a constant level. In a power supply that happens over an over at switching frequency of the regulator gating the mosfets on/of and is the basis of a DC power supply filter capacitor.

When caps are used in bypass and DC de-coupling/AC coupling circuits the DC component they see is already filtered and in a steady state. These caps charge to the DC voltage level applied to them at power up and block DC current flow after that but will still allow the lower AC voltage level signals to pass through.

I guess you could call it conducting the AC signal component but that would be rather confusing since the DC component is blocked once the cap is charged to the DC voltage level. That charge time only takes a couple of milli-seconds at power up and is what separates capacitance from conductance.

Viper