Replaced capacitor on gtx 970

[Thegovahnator]

Last week i broke off a capacitor of my MSI gtx970 4g.
it's a 16v 270uf capacitor.

I have replaced it with a 270uf cap (voltage is not noted on the cap so i am not sure about that) taken from a broken p67 pro3 se mb.
The capacitor is physically about 1/3rd taller.

Now the card works fine as long as i limit the TDP in afterburner to +-75 percent. It still runs at 1310 mhz turbo boost just like on 100 percent TDP.
If i run it on full power, the TDP in gpu-z rises really fast and when it reaches 100 percent the pc switches off.

Is the card acting like there is no capacitor installed (maybe it doesn't make contact like it should evne though it looks fine), or does the card need less power to do the same with this capacitor?



Anyway, i am trying to change the TDP in bios to 75 percent with lowering values to represent +-75percent , but it doesn't seem to work.
Maybe someone can tell me exacly what to change to achieve this? I already followed a maxwell guide but after flashing the bios it just acts the same as before.

 

[SkyNetRising]

If i run it on full power, the TDP in gpu-z rises really fast and when it reaches 100 percent the pc switches off.

That sounds like PSU protection kicking in. What's the model name of your PSU?

 

[Thegovahnator]

If i run it on full power, the TDP in gpu-z rises really fast and when it reaches 100 percent the pc switches off.

That sounds like PSU protection kicking in. What's the model name of your PSU?

It is, asus surge protection kicks in.
It's because the card takes to much power since the cap fell off/was replaced.

ocz stealth extreme 2 600w

 

[jasonkaler]

Did you test the new capacitor you installed?

There are a few things that could be happening, just off the top of my head:
1) the new cap you installed isn't able to hold that full capacity
Caps degrade with time, or it could be a lower voltage, even if it is physically bigger
2) the new cap isn't filtering out the correct frequencies.
Caps can also used as filters and the designer would be choosing their caps based on the specific frequency range it is filtering. Different substances within the cap have big effects here
3) there is now a short, or the new solder is running too close to a trace.

Can you post some pics?

 

[Thegovahnator]

Did you test the new capacitor you installed?

There are a few things that could be happening, just off the top of my head:
1) the new cap you installed isn't able to hold that full capacity
Caps degrade with time, or it could be a lower voltage, even if it is physically bigger
2) the new cap isn't filtering out the correct frequencies.
Caps can also used as filters and the designer would be choosing their caps based on the specific frequency range it is filtering. Different substances within the cap have big effects here
3) there is now a short, or the new solder is running too close to a trace.



Can you post some pics?

Thanks for the answer, i did not test the cap.

I will be home in a few hours and will take some pics.

So i guess you agree that the cap is making contact (mallfuntioning though). It is not possible that it works fine; but that it just needs less power to perform the same?

I am quite proud how i managed to replace the cap, the soldering is allmost done as good as it was original .

upload pictures online

I was thinking aswell that it might be a 11v cap that i placed.

I have another broken motherboard around, with also a taller cap but one that is 16v 270uf for sure. I didn't want to take it because i am planning to bring in the board into a shop since it still semi works.

 

[jasonkaler]

My best guess is that the cap is connected but just not storing enough charge.
Then, when the power consumption goes too high, there is not enough energy left in the cap and it pulls too much current from the PSU, causing the psu to shut off that rail.

I'm also assuming you put the cap the right way round. I'm pretty sure you would have seen some nice smoke if you got it wrong.

 

[Thegovahnator]

My best guess is that the cap is connected but just not storing enough charge.
Then, when the power consumption goes too high, there is not enough energy left in the cap and it pulls too much current from the PSU, causing the psu to shut off that rail.

I'm also assuming you put the cap the right way round. I'm pretty sure you would have seen some nice smoke if you got it wrong.

I forgot to take a picture, but i replaced the cap with a different one from an asus p67 motherboard.
It is a 16v 270 uf 11A cap. Maybe the 11A is what it needs. The one i had before had 0ya marked.

I still need to test it on power hungry programs to check if i have the same issue with this one, but it booted without issue.

 

[Thegovahnator]

So, i tested with a the different cap.

All worked well, but after closing battlefield 1 the pc fell out.
And it won't boot now when the pcie1 connector is plugged in the gtx 970 :s
My gtx 660 works fine

 

[jasonkaler]

ouch.
hopefully there's no physical damage or shorts.

 

[Thegovahnator]

ouch.
hopefully there's no physical damage or shorts.

I removed the cap, and it booted again . But i turned it off immediately.

I will try with even another cap i guess.

I found caps with 0ya,0yd and 11A (on same donor board) all 16v and 270uf. Any idea which one i should take and what the difference is?

 

[jasonkaler]

Codes like that are heat or capacitance tolerance.
A = +- 1% capacitance change over entire temperature rating - A is the best possible for high temperature use.
D = +- 3.3%
Y = min temp: -30 degrees C

http://www.robotoid.com/appnotes/electronics-capacitor-markings.html

As the card probably only runs at 12v, I don't really see how such small differences will have any effect as the cap is rated at 16v, which is a whole 33% above what you require. The caps with "A" are probably better quality though.

The cap might also have its rated temperature. That's probably more important. The board infront of me has most caps at 85deg, but the one right next to the voltage regulator is 105deg.
You probably want to use a Y6A or similar, but the 16v 270uf is not negotiable.

 

[RobCrezz]

Codes like that are heat or capacitance tolerance.
A = +- 1% capacitance change over entire temperature rating - A is the best possible for high temperature use.
D = +- 3.3%
Y = min temp: -30 degrees C

http://www.robotoid.com/appnotes/electronics-capacitor-markings.html

As the card probably only runs at 12v, I don't really see how such small differences will have any effect as the cap is rated at 16v, which is a whole 33% above what you require. The caps with "A" are probably better quality though.

The cap might also have its rated temperature. That's probably more important. The board infront of me has most caps at 85deg, but the one right next to the voltage regulator is 105deg.
You probably want to use a Y6A or similar, but the 16v 270uf is not negotiable.

Pretty sure other stuff on the card runs a different voltage to 12v, the 12v feeds have to be transformed (GPU for example runs at around 1v, the vram 1.55v etc.

 

[jasonkaler]

Pretty sure other stuff on the card runs a different voltage to 12v, the 12v feeds have to be transformed (GPU for example runs at around 1v, the vram 1.55v etc.

correct. Each section of the card will probably have it's own specified voltage.
Those large caps are usually near a regulator or buck converter, dealing with the voltage coming directly from the PSU. The voltage gets stepped down to the correct voltage by the buck converter and the capacitor does two things:
1: store charge for the converter
2: reduce noise.
The cap works as a local reserve of power, allowing components further down to draw large amounts power in short bursts. These quick pulls then come from the cap rather than from the PSU reducing spikes at the PSU and other components on the same rail.
You will also find smaller caps after the regulator, which will handle the lower voltages. These would most likely be surface mount as the voltage requirements are small.