[SOLVED] I carelessly broke off a VRM chip.

Squatch82

Commendable
Apr 8, 2017
8
0
1,510
Today I had disassembled my GTX 1070 FE to clean it up and I carelessly broke off what I understand to be a VRM for the memory. The spot where the chip belongs is labeled C127. I promptly reassembled the card and loaded up Valley benchmark to see if my card was now dead and everything seems to be running fine. Does anybody know if my card will be ok or is it now dying a slow death (redacted)?
20190723_085315.jpg
 
Solution
Oh, you broke the MOSFET...
That's a capacitor, not a MOSFET. It's likely in parallel with all the other caps (probably a VRM output cap), so basically all that's happened is the total output capacitance has been reduced by 1/X relative to what it was, where X is the total number of caps in parallel. E.g. if there were 12 caps now the output capacitance is 11/12 of what it was. Could have some impact on stability, but probably not a big deal.

VRMs work with a lot of current, so if they arch, bad things will happen. It may be working now and even normal (non-stressed) conditions, but if you put too much current, they will get hot and an arch is bound to form at some point.
The highest voltage on the board is 12V. I'm...
If you just broke the "pin heads" from the other side of the lil' cockroach then you're safe. If you broke the connecting metal to the front side of the PCB, then that's a fire hazard waiting to happen IMO. VRMs work with a lot of current, so if they arch, bad things will happen. It may be working now and even normal (non-stressed) conditions, but if you put too much current, they will get hot and an arch is bound to form at some point.

Can you post better images as well? That image is hardly helpful with the bad drawings on top.

Cheers!
 
Oh, you broke the MOSFET...

The card will keep working alright, but it will freeze when that power rail/phase is used and/if the MOSFET is actually broken (I can't really see if you took part of it with the cap or the cap itself is the whole piece).

Theoretically you can replace them if you find the exact same component and you have good soldering skills, but for 90% of people out there I would suggest against it.

The card will keep on working, as far as I'm aware, but if you OC or the card pushes itself too hard, you will get random behaviour and, probably, damage the GPU itself. If you still have warranty left, try using it?

Cheers!
 

Squatch82

Commendable
Apr 8, 2017
8
0
1,510
Thank you for the reply. Unfortunately I have no warranty as I bought it second hand. So you are saying if I just run it at stock clock speeds I should be fine? And what did you mean by taking part of it with the cap or the whole cap itself?
 
Thank you for the reply. Unfortunately I have no warranty as I bought it second hand. So you are saying if I just run it at stock clock speeds I should be fine? And what did you mean by taking part of it with the cap or the whole cap itself?
Well, some manufacturers use "caps" or "wraps" on certain components to either protect them or give them some volume and they don't really serve any practical purpose. From the image alone, I can't really make if this plastic you removed is just a protective layer (ideal-ish scenario) or it's part of the component itself (busted scenario).

And running stock does not mean you won't run into issues. Remember this is part of the power delivery circuitry, so it will depend on how the power controller splits the VRMs for power and whatnot. I have zero idea how it is for your card, but what I said should be true: as long as that particular rail is not used, you won't have problems. Chances of* that, at high stress, are low.

Cheers!
 
Last edited:

Squatch82

Commendable
Apr 8, 2017
8
0
1,510
Ok so all of that being said, the issues I may have will just be performance and the way the card behaves while gaming and whatnot? We aren't talking something like a fire hazard right? I know you said the before when I thought it was the vrm and not the mosfet. Thank you so much for all of your help!
 
Ok so all of that being said, the issues I may have will just be performance and the way the card behaves while gaming and whatnot? We aren't talking something like a fire hazard right? I know you said the before when I thought it was the vrm and not the mosfet. Thank you so much for all of your help!
As each MOSFET is susceptible to burning anyway, no. Ironically enough, if you push the power envelope way too much, they just go "pop". Maybe that plastic is for exactly that purpose: so it doesn't cause a fire hazard in case it blows and has a non-conductive layer preventing archs?

You would need to grab a more detailed guide on MOSFETs, as my knowledge stops there, I'm afraid. From here on out, it will be up to your risk adversity whether or not you want to risk it. I would try to glue the cap anyway, as it should still provide cover if the MOSFET is indeed busted. Plus, it did not appear it was soldered at all, so chances it's just a "wrap" are not low.

Cheers!
 

TJ Hooker

Titan
Ambassador
Oh, you broke the MOSFET...
That's a capacitor, not a MOSFET. It's likely in parallel with all the other caps (probably a VRM output cap), so basically all that's happened is the total output capacitance has been reduced by 1/X relative to what it was, where X is the total number of caps in parallel. E.g. if there were 12 caps now the output capacitance is 11/12 of what it was. Could have some impact on stability, but probably not a big deal.

VRMs work with a lot of current, so if they arch, bad things will happen. It may be working now and even normal (non-stressed) conditions, but if you put too much current, they will get hot and an arch is bound to form at some point.
The highest voltage on the board is 12V. I'm pretty sure those are low side capacitors, so they'd only have ~1.x volts across them. It's not going to arc from that kind of voltage.
 
Last edited:
  • Like
Reactions: Squatch82
Solution
That's a capacitor, not a MOSFET. It's likely in parallel with all the other caps (probably a VRM output cap), so basically all that's happened is the total output capacitance has been reduced by 1/X relative to what it was, where X is the total number of caps in parallel. E.g. if there were 12 caps now the output capacitance is 11/12 of what it was. Could have some impact on stability, but probably not a big deal.
I'm pretty sure it's not a capacitor... I could buy you saying it's a diode, but not a capacitor... Are you 100% sure?

The highest voltage on the board is 12V. I'm pretty sure those are low side capacitors, so they'd only have ~1.x volts across them. It's not going to arc from that kind of voltage.
Voltage doesn't kill; Amps (current) do.

And that's exactly the problem I'm pointing at: it will depend on what is the max current that would transit through it if it's dangerous or not.

Cheers!
 

TJ Hooker

Titan
Ambassador
I'm pretty sure it's not a capacitor... I could buy you saying it's a diode, but not a capacitor... Are you 100% sure?
Yes. An easy way to tell is that it's labelled C127. Capacitors are labeled starting with 'C'. Diode labels would start with 'D', transistors (FETs) with 'Q', resistors with 'R', etc. You can also tell it's definitely not a transistor because it only has two terminals.

Voltage doesn't kill; Amps (current) do.

And that's exactly the problem I'm pointing at: it will depend on what is the max current that would transit through it if it's dangerous or not.
Either voltage excessive voltage or current can damage a component. But we're not talking about damaging the component, because the component is gone...
You said there could be arcing. Arcing is 100% a result of voltage. It's not clear to me exactly what/were you thought arcing would occur, but given the low voltages involved I can't see it happening anywhere.
 
Yes. An easy way to tell is that it's labelled C127. Capacitors are labeled starting with 'C'. Diode labels would start with 'D', transistors (FETs) with 'Q', resistors with 'R', etc. You can also tell it's definitely not a transistor because it only has two terminals.
Ok. That makes perfect sense to me. So the plastic thingy is not part of the capacitor itself then?

Either voltage excessive voltage or current can damage a component. But we're not talking about damaging the component, because the component is gone...
You said there could be arcing. Arcing is 100% a result of voltage. It's not clear to me exactly what/were you thought arcing would occur, but given the low voltages involved I can't see it happening anywhere.
You're not wrong about the arch being a product of a potential difference (volts), but it takes very little voltage to create an arch over 2 conductive materials in a tight space as breaking the resistance of air in tight spaces is easier. Otherwise, we wouldn't be seeing things like the GTX590!

View: https://youtu.be/sRo-1VFMcbc?t=26


Looks at how it sparkles! The shinnies! Well, I don't know if it ached anywhere, but given the light you can see from inside, it must have.

--

Anyway, OP, that's what happens when you basically aren't careful with these things.

Cheers!
 

TJ Hooker

Titan
Ambassador
I'm not sure what "plastic thingy" you're referring to. Pictures aren't very clear though.

It's a little different when you capture a component in the act of failing. The flash of light could just be the component burning up. If there is arcing, it could be a result of a microscopic void/failure forming in the component, such that even a small voltage can create a very high electric field due to the tiny distances involved. If we're talking about macroscopic distances e.g. on the order of mm you're going to need a lot higher voltage for air to breakdown.
 
Last edited:

Squatch82

Commendable
Apr 8, 2017
8
0
1,510
So I've done some digging around and it turns out that it is indeed a capacitor. Hopefully it doesnt create a hazard. I dont have the money to replace the card right now so I guess i will just run it and hope that it continues to work for the time being.