Adding Capacitors to GPU VRMs Beneficial?

[Dayaks]

Has anyone ever added capacitors to a GPUs VRMs?

Long story short, was modding a card and ripped off one of those surface mount caps destroying it in the process. I soldered on a relatively large electrolytic cap (330 uF). Seemed I could do +50 on the core. I didn't really do a super great job before and after testing.

Anyone ever bother with this? Makes me wonder about adding capacitance to reduce ripple. I also noticed there's surface mount caps on the back of the board directly behind the chip.

 

[cageymaru]

Would you mind posting pictures of it? I love seeing stuff that people fixed themselves. How hard was it to fix?

 

[EchoWars]

Would not have been tough or expensive to get the proper cap and solder it in.

 

[Dayaks]

Would you mind posting pictures of it? I love seeing stuff that people fixed themselves. How hard was it to fix?

It was very easy to fix. Spent more time getting the cap out of the drawer and waiting for the iron to heat up.

I won't be getting pictures for some time. My computer is drilled into the desk so the kids/wife can't get at it.

Would not have been tough or expensive to get the proper cap and solder it in.

I couldn't see a negative to using a larger one? Hell it seems I got an extra 50 MHz out of it... It's obviously used for ripple reduction.

It actually would have been pretty hard to solder the surface mount cap due to it's physical orientation anyways.


I was just curious if anyone has messed around with this. We do hard power limit mods, bios mods and custom loops. It's not anymore difficult than any of those. Maybe on my next business trip I'll do it to my 970. Worst case I break it and have to grab a 1070.

 

[Azrak]

I won't be getting pictures for some time. My computer is drilled into the desk so the kids/wife can't get at it.

I couldn't see a negative to using a larger one?

Capacitors are chosen not only for their actual capacitance (easy to meet) and voltage (also easy to meet), but also ESR (equivalent series resistance) and temperature rating (which is used to get the lifetime hours). Ripple current rating is also very important for power circuit caps. All of these together are harder to match in a replacement.
That's why replacement caps for motherboards are a PITA to source sometimes because the ESR has to be low (and most are not), ripple current has to be high (and matched if possible) and often the dead caps belong to a long-gone series that is no longer manufactured or was known for bad electrolyte.

Larger capacitance caps don't do as well with faster voltage changes (higher frequency ripple), and the additional ESR they have (compared to smaller valued caps) can load the power circuitry in ways that may not be as good as the circuit wants to have to stay in spec.

So yeah, short term, the cap you replaced will work (I mean, it's working now, right?), long term though you are probably stressing other power circuitry components due to the larger cap value and different ESR and ripple current ratings from spec. But without knowing the circuit design and critical/non-critical nature of the specific capacitor you replaced though, nobody will ever know, at least until something dies. And it may never die. That's the fun thing about electronics that are hacked. It will work for a while and poof out. And the length of "a while" could be years or days.

Hopefully you used a 105 C rated cap.

 

[Dayaks]

Capacitors are chosen not only for their actual capacitance (easy to meet) and voltage (also easy to meet), but also ESR (equivalent series resistance) and temperature rating (which is used to get the lifetime hours). Ripple current rating is also very important for power circuit caps. All of these together are harder to match in a replacement.
That's why replacement caps for motherboards are a PITA to source sometimes because the ESR has to be low (and most are not), ripple current has to be high (and matched if possible) and often the dead caps belong to a long-gone series that is no longer manufactured or was known for bad electrolyte.

Larger capacitance caps don't do as well with faster voltage changes (higher frequency ripple), and the additional ESR they have (compared to smaller valued caps) can load the power circuitry in ways that may not be as good as the circuit wants to have to stay in spec.

So yeah, short term, the cap you replaced will work (I mean, it's working now, right?), long term though you are probably stressing other power circuitry components due to the larger cap value and different ESR and ripple current ratings from spec. But without knowing the circuit design and critical/non-critical nature of the specific capacitor you replaced though, nobody will ever know, at least until something dies. And it may never die. That's the fun thing about electronics that are hacked. It will work for a while and poof out. And the length of "a while" could be years or days.

Hopefully you used a 105 C rated cap.

I did think about current since that cap I ripped off did have quite thick wire.... or thin wide flat wire. My OP was more aimed at supplementing the stock caps so it's less risky. Maybe large ceramic caps would be better. I've never heard of those blowing up.

So sounds like it's best to buy those crazy AIB boards that come with relative massive caps and phases.

Or if you want to DIY try to buy the caps on the board and parallel them. Still changing things of course, but probably so is running our 250W boards at 400W. There's a reason I bought a mostly metal case.

You are right these can be hard to find. Not a whole lot of info on them. I did give a half assed try before going into my parts drawer.

All good points. Nice reply.

 

[Vudaz]

It's there any chance to fried your card? I mean if something goes wrong?

Thanks

 

[Dayaks]

It's there any chance to fried your card? I mean if something goes wrong?

Thanks

To Azrak's point if you pick the wrong capacitor, or if the capacitor is polarized (has a + and -, electrolytic are like this), you can hook it up backwards and catch it on fire.

Ceramics should be pretty safe.... everything has a risk and you would be voiding the warranty (which I personally do anyways).

 

[Vudaz]

To Azrak's point if you pick the wrong capacitor, or if the capacitor is polarized (has a + and -, electrolytic are like this), you can hook it up backwards and catch it on fire.

Ceramics should be pretty safe.... everything has a risk and you would be voiding the warranty (which I personally do anyways).

I was asking myself what would be the benefits x risks. Is there a significant increase of power o justified such an art?

thanks

 

[Dayaks]

I was asking myself what would be the benefits x risks. Is there a significant increase of power o justified such an art?

thanks

Significant increase? I really doubt it.

Justified? It's more for the fun of it.

Again I think I saw a small, basically insignificant increase. I made the OP just wondering if anyone else has dabbled.

 

[Vudaz]

Significant increase? I really doubt it.

Justified? It's more for the fun of it.

Again I think I saw a small, basically insignificant increase. I made the OP just wondering if anyone else has dabbled.

I got the point. I love fix things. I would like to know how to measure the circuits and try to bring back my old 8800gt...

 

[Deleted member 93354]

Has anyone ever added capacitors to a GPUs VRMs?

Long story short, was modding a card and ripped off one of those surface mount caps destroying it in the process. I soldered on a relatively large electrolytic cap (330 uF). Seemed I could do +50 on the core. I didn't really do a super great job before and after testing.

Anyone ever bother with this? Makes me wonder about adding capacitance to reduce ripple. I also noticed there's surface mount caps on the back of the board directly behind the chip.

Dear god No. Don't even. You could overload the output on the VRM. Remember when capacitors are discharged completely they act like a ground source because all the energy goes there first to power them up. If the VRM isn't up to snuff current wise on supplying both the capacitor and component requirement you get a FAIL.

 

[Dayaks]

Dear god No. Don't even. You could overload the output on the VRM. Remember when capacitors are discharged completely they act like a ground source because all the energy goes there first to power them up. If the VRM isn't up to snuff current wise on supplying both the capacitor and component requirement you get a FAIL.

Shouldn't the inductor protect the FETs?

 

[Deleted member 93354]

Shouldn't the inductor protect the FETs?

An inductor controls current at the expense of voltage. However is the capacitor is taking 90% of that then the remaining components will fail due to not enough current which will cause a voltage drop.

Larger caps can help with transient voltage changes. However a lot of caps can create ripple as the F goes up as the variation in the charge surface increases with the cap surface area increase. Larger Cap = larger plates = larger variations = more ripple. A quality cap can help with that. But I wouldn't mess with it if you aren't sure.