Please help me sort out my v-mod accident

[Revivalist]

Well, this last weekend I had a very tragic volt-mod accident trying to mod a 7800GTX following this guide. I want to learn from my mistake because I really don't want to give up this part of the hobby. I know I can do it right if I just do it carefully and understand all the facts.

After doing lots of research on the net, I came across this example in our forums and also this example from XS forums. The second looks like this . . .

I basically followed this example and I included a female molex soldered to the voltage measuring points so that I could easily measure the vGPU and VDD voltages while the card is running. Since the people in the links above seemed to have success, I figured I would follow them and use Viper John's schematic as a reference. (I don't have a pictures of my card handy but it ended up looking similar.)

But when I plugged in my card and fired up the computer, I smelled a small hint of smoke. So I turned off the system right away and checked that the trim pots were turned for maximum resistance. Then I put the probe of my volt-meter into the female molex to check the vGPU. But when I turned it on again, a huge puff of smoke came out of my case. . . .

I’m trying to figure out what went wrong so I can avoid this in the future. But I'm having trouble figuring it out. I have a couple questions that I was hoping someone could answer to help clarify the problem. . . .

1. If I solder the ends of a trim pot to the card, then can I measure the resistance between the soldered points to know the resistance of the trim pot? . . . When I did the v-mod on the GPU, I first checked my 200 ohm trimmer was set to 200 ohms between the two prongs that I soldered (the two on the right). I even snipped off the one on the far left to avoid confusion or accidental contact. Then I soldered to the points shown in the guide. But when I put my meter to check the resistance between the two points I soldered, it was something other than 200 ohms (I don’t remember exactly but I think I remember something like 38 ohms which I know means major current. ) . . .. So the resistance between the soldered points should have been around 200 ohms, right? And if so, why would it be so much lower in my case?

2. Does it matter where on the card I use as a ground? . . . In the guide, Viper John used some capacitors as ground points because they were near the v-mod points. However, since I like to put my trimmer at the edge of the video card (not directly located at the v-mod point) I figured I would just use the ground located by the power connector (like in the picture above). Would this be a problem? (I didn’t think so because this person seems to have done it and worked just fine.) . . .

3. Does it matter if the volt-meter is grounded to the case when reading voltages? . . . At first I thought that was the reason the card smoked the second time I turned it on. . . . But thinking about it later, that didn’t really make sense. The volt-meter doesn’t act as a path to ground, right? It shouldn’t matter if the volt-meter is grounded on the card or on the case, true?

Thanks for any help, guys.

 

[crimedog]

VJ's mods don't include the ground at all. you solder the pot between two points on the pcb. you use different points if you want to do to ground.

i assume you mixed the two different volt mods? i am having a very hard time figuring out what you did.

 

[four4875]

the resistance showed differently when the pot was soldered on the card because the card is also allowing the current to pass, so its going through both the pot and the card.

you should be able to use any ground, as long as it is in fact a ground.

the case should be physically connected to the ground of the mobo, vid card, every ground in the system. that shouldnt have been a problem.

now you get to figure out where the smoke came from, and replace that and revive your card. or at least try to. what do ya have to lose? . hopefully it was just a little resistor or a vreg chip or somethin that will replace relatively easily, and you can spend a couple $ and have your card working again.

when you set the pots to max resistance, the multimeter read --- or whatever, not 0 right? 0 ohms is no resistance, which would probably be problematic lol. and if you're sure all the connections are right (and that your card didnt happen to have a tiny variation in design and fooled ya or somethin) and the pots were set correctly... i have no idea what coulda went wrong.

i think that the problem may have been the pot setting at 200 ohms. while i havnt read the mod guides, all mods ive seen you should start with infinite resistance and slowly drop it till you get a change in voltage, to prevent from jumping straight in with too much voltage.

one last thing i can think of that would have caused problems would be if you had the multimeter set to current instead of voltage, it would have given the card a nice nearly dead short to ground, which would be very bad.

 

[four4875]

VJ's mods don't include the ground at all. you solder the pot between two points on the pcb. you use different points if you want to do to ground.

i assume you mixed the two different volt mods? i am having a very hard time figuring out what you did.

uhh... yeah if you followed the first guide, the one you linked and said you used.. i dont think there should have been any grounding of pots. that could be VERY problematic lol.

 

[Revivalist]

Thank you very, very much for the help, guys! . . . I'm sorry if what I explained sounds confusing. I'm trying to explain it as best as possible. Just let me know what is unclear.

I might seem crazy, but I'm going to get a replacement card and attempt this again on Thursday or Friday. I'm also working on a "video card v-mod" guide that I'm hoping will help others to avoid the mistakes that I've made. So I really appreciate the help. . . .

As for the VJ mods, I'm pretty sure that it does include points that go to ground. The points that are soldered on capacitors are grounded and the points that are soldered on "pads" on the PCB are the v-mod points. . . . Check out posts #369-374 in the 7800GTX v-mod thread at XS Forums. You can see that someone used the grounded points at the capacitors and also the ground at the 6-pin connector. They ended up connecting grounds all around the card. (That's funny! ) . . . But notice also that they fixed it by connecting the v-mod points to the ground at the 6-pin connector. That worked fine. It's the photo that I posted at the top of this thread.

The reason VJ uses those capacitors for ground points is because they are close to the v-mod points and he likes to stick the trimmer right at the location where the mod is done in the middle of the card. (He says that this reduces the EMI from wires running along the card.) But for someone who puts the trimmers on the edge of the card (like me), it is not necessary to use the ground points at the those capacitors. In fact, I figured it would be better to use the ground near the 6-pin power connector location because it is near the edge of the card so there would be less wires running along the card. . . .

Thank you for the response to all my questions. That makes sense that the resistance showed differently when the pot was soldered on the card because the card is also allowing the current to pass. I should of thought of that. . . .

As for figuring out what burned up, I'll give it a try, but I'm not very hopeful about finding it and being able to replace it. . . .

It was mentioned that I should start with infinite resistance and slowly drop it till I get a change in voltage. But how would I start with infinite resitance? I have to start with the maximum resistance that the trimmer allows, right? I figured that VJ and whoever else figured out the v-mod, are suggesting trimmers that basically leave the voltage unchanged when they are at they're max so that they can soldered on and then adjusted to slowly increase the voltage. Is that right?

 

[SuperFarStucker]

wow, that's a bummer, I was half expecting that to happen when I did my mod. I'm not exactly an expert on the subject but I calibrated the potentiometers before I hooked them up by measuring the resistance across them with my voltmeter. It sounds like you did the same thing. Some of the points are really close together for the vgpu, did you make sure you didn't bridge any of them? It really helps to have magnification, it took me several tries to get that one right. It also sounds like you tried to combine two different mods and that won't necessarily work. Sorry to hear about the loss but as others have suggested perhaps it is just a cap or resistor which could be fixed.

 

[crimedog]

i once pm'ed VJ about whether any of those points were ground, because i thought the pads sucked to solder to, and he said they were not. please dont kill another card.

 

[Revivalist]

@ SuperFarStucker - Thanks for the condolences. Yeah, I did make sure to calibrate the pots before soldering them. I don't think I accidentally bridged any connections but it's certainly possible. . . . The v-mod I did is exactly like the one in the picture at the top of this thread. I figured it was the same as VJ's guide with the only difference being the ground points. It seemed to work for the person that took that picture. So I don't know. . . . But I don't think I'll risk doing it that way again.

@ crimedog - Wow, that's interesting! VJ said that none of those points are grounded? The pads that you mentioned are the v-mod points so I know those can't be avoided even though they're difficult to solder to. But I thought that soldering to those capacitors near the pads is not necessary because they are one of many ground points that could be used. . . . Hmmm, or maybe I'm not really understanding what's being said in this thread I posted. For example, in this post, VJ says, "I always pick up the ground plane close to the mod point to keep noise down in the regulator feedback circuit." I understood that to mean the there is a grounded point at each of the v-mod locations. Did you happen to look at posts #369-374 in that thread as well? It seems that someone actually soldered only to the grounded points that VJ gave as well as to the ground at the 6-pin connector. That's why they basically connected ground points around the video card. But when he resoldered to the v-mod points it worked. . . .

Anyways, I think I'll take your advice and just follow VJ's guide exactly in order to avoid killing another card. Believe me, I really don't want to repeat that little experience. . . .

One last thing . . . . what do you guys think about putting the pots at the edge of the board? VJ suggests not doing so because it feeds EMI right into the voltage regulator underneath the card. However, I had a bad experience in the past killing a card by trying to measure a voltage and accidentally touching two points on the card and shorting something. So I thought it seems too dangerous to put a pot in the middle of the video card and stick a screw driver in there to adjust. I guess I could turn off the computer every time I adjust the pot value but that would be a real pain! . . . . Did you guys put your pots on the edge of the card or right at the v-mod location?

 

[ThePCGuy]

Hey, umm, don't throw away that card . I'm sure someone on here could try and fix it. Especially being that it's a 7800 . I was just skimming and wasn't sure if it was going to get tossed. If you decide you want to trash it, send it to me if you will . But then again I don't know if it's AGP or PCI-E as I only have AGP. Eitherway, it's late and I figured I'd just pop in.

- Jim

 

[crimedog]

i have seen alternate mod points that go to ground, but they were most certainly not the same as VJ's points hosted on malvescorner.net. furthermore i wouldn't trust some kid being made fun of on xs who says it's his first time soldering either

personally i put my pots on the edge of the card, along with a female molex for read points. i did it that way and it works great for me.

edit: i guess my memory sucks. i searched my history and VJ said that the caps you solder to were infact ground. guess something just wasn't kosher.

 

[Revivalist]

@ ThePCGuy - No, I wasn't going to throw the card away. I think it can be fixed. However, if I decide to sell it cheap, you'll be the first to know. (But it is a PCI-E card.)

@ crimedog - Yeah, I agree about not trusting someone's work who is not experienced. But others in the thread confirmed that he did it right and he reported his results. So it seems it's legit. . . .

Thanks for checking on that pm from VJ. . . . I thought I was going crazy there for a minutre, either that or I was totally misunderstanding what people were saying in that thread at XS. . . .

But now this really leaves me wondering what I did wrong? . . . There was one other deviation from VJ's guide that I did. . . . For the vGPU read point, I used the same location that "bazx" uses in this thread instead of the one VJ pointed out. It's measured at that chip that's just below the screw for the RAM heatsink bracket. (I checked that location before I did the mod and it was in fact giving the the proper vGPU value.) I chose that location because the location VJ points out is really close to the GPU (right between the GPU heatsink bracket and the RAM heatsink bracket.) That point would actually be covered by the neoprene insulation I plan to use for a pelt setup. Should I just use that location VJ uses and let the neoprene cover it? Or should I follow the example of bazx? . . .

By the way, did you mean to put "(1.67v, 2.35v)" in your sig, not "(1.67v, 1.35v)"?

 

[crimedog]

if a read point works it works, i used VJ's and just did the mod extremely low profile since it was being covered with a backplate and I didn't want it sticking up. yes you're correct about 1.35v, thanks for pointing that out

 

[Revivalist]

Yeah, I like to make mine very low profile as well to avoid wires accidentally getting pulled later. In fact, I just conformal coat them down . . .

Well I guess since things didn't work out the first time, I'm going to try this next time by just following VJ's guide exactly. Hopefully I can't go wrong that way. . . I'll just cut the neoprene around the vGPU measuring point. I don't think I'll put the pad right over it in case I need to remove the pad later.