[GUIDE] GTX 1080 Resistor Power Mod

[graxxmusic]

This is a guide on how to raise the target power limit of your GTX 1080 with a resistor mod. In my case, I used an EVGA GTX 1080 FE. Your card can differ from mine.
I am new to forums (actually this is my first informative post) so I hope it'll help you. Feedback always welcome! 

After doing some research about modding the shunt resistor on GTX video cards to raise power limits, I decided to do skip the liquid metal part because the composition of this metal could damage the resistor/tin and there is also a chance that the liquid metal could drip off the shunt and short other components. But beware that this is basically a permanent mod and can hardly be reversed. Also, some soldering and electrical skills needed!

  

Figure 1

What I did is soldering another resistor parallelly on top of the shunt resistor. The reason for soldering a resistor parallelly instead of replacing is because we are working with a very low resistance (1mΩ = 0,001Ω), and putting your own calculated resistor with too much soldering tin will create an even bigger resistance where you can't know the exact resistance. Here you can see why:

  

 Figure 2 & 3

The two bigger pads are where the current is mostly running trough and two tiny pads that are connected to the power measurement circuit. It's important to keep the resistor as close as possible to the sensing pads for a more accurate measure.
I hear you thinking: but you are also raising the resistance if your soldering a resistor on top of it, isn't that the same? Not exactly, because if your value is not correct you can carefully solder it off and replace it with another value (I recommend to calculate it right at once). in brief, It's definitely an important aspect to think of if you are working with low resistance's.

As you can see in figure 1 I soldered a 6mΩ 1% resistor on top of the existing 2mΩ resistor, raising the card's power limit by 33%. I will explain how I calculated that.

I used the common formula for calculating parallel resistance :

 Figure 4

In my case, I wanted to raise a maximum of 33% of the power limit.

That's going to be: 2mOhm / 100% * 25% = 0.5mOhm subtracting that from 2mOhm = an Rtot of atleast 1,5mOhm.
Subtracting 25% from the resistor increases the power limit by 33%:

2 mOhm / 1,5 mOhm * 100% = 133%

R1 = 2mΩ
Rtot = 1,5mΩ 
We want to know R2 so the formula becomes:

 

 = 6mΩ

 Figure 5

After that, I assembled the card back together and tested it with Furmark.

Before the mod:

 

Figure 6

After the mod:

Figure 7

As you can see I gained a fair amount of core clock, and also the VDDC has more room to increase. I actually did need to lower the core overclock because it was running too fast and began to crash.
The tests were running quite short which is why the card was not running hot. 
But I'm not sensing any extreme temperatures on the longer runs (30 mins on 100% fan speed), but this could be that I used liquid metal on the GPU die:

 

Figure 8 & 9

Now, you can extend it further than 33% but be aware that your card can run very hot or even blow up. So if you are planning to do this, calculate the resistance well, use a more extreme cooling solution (like custom water cooling) and keep a good eye on the temps so that the risks are minimalized.
After all, it's a limit to protect the card from breaking. Do it at your own risk.

Thank you for your time, I hope it was helpful. I will start shortly on my first hardline water cooling build which I will post pictures of too.

 

Yours sincerely,

 

Graxx

 

Sorry for my bad English, I'm dutch.

 

 

My sources:

Google is your best friend

der8auer's video:

 

Edited February 12, 2020 by graxxmusic

[evassion]

Hello,  one fast question!

How can I know the shunt orientation in the pcb before soldering? 

I did not take a photo before "liquid metaling" them and they fell after two years. 

My plan it is to do like this and use 3Momh directly (the pcie shunt is untouched) i did a photo:

 

https://ibb.co/jDCK21p

 

Thanks! 

Edited June 16, 2019 by evassion
Correction

[niobium615]

They're just resistors, so the orientation doesn't matter.  As long as both pads are connected, you're good to go.

[Trendle]

Doesn't this increase the power limit by 33%, though? -25% resistance on 2 mohm = 1.5 mohm, correct. But, 2 / 1.5 mohm = 1.33, so 217 W (EVGA GTX 1080 FE power limit) * 1.33 = 289 W, or 33% increase over BIOS limit. Right?

[graxxmusic]

 

On 2/6/2020 at 11:46 PM, Trendle said:

Doesn't this increase the power limit by 33%, though? -25% resistance on 2 mohm = 1.5 mohm, correct. But, 2 / 1.5 mohm = 1.33, so 217 W (EVGA GTX 1080 FE power limit) * 1.33 = 289 W, or 33% increase over BIOS limit. Right?

Hello!

As you confirmed, 25% of 2mOhm is 1.5 mOhm. I indeed made a mistake to relate the 25% change in resistance with the output power.
I had to think of that the power (W), is relative to current (I) or voltage (U) because the calculation is squared (²).
This is what I should've done:

Example fixed value:
U = 12V

Situation 1:
R = 2 mOhm
P = U² / R so, 12V² / 2 mOhm = 72000W

Situation 2
R = 1,5 mOhm
P = U² / R so, 12V² / 1.5 mOhm = 96000W

Difference 24000W

100 * 24000 / 72000W = 33% increase.

Thank you so much for the heads up, I'm going to change it to 33%.

P.S. Maybe my calculation here is obsolete because you probably understand it quite correct, but yeah why not 
 

[graxxmusic]

On 6/16/2019 at 11:53 PM, evassion said:

Hello,  one fast question!

How can I know the shunt orientation in the pcb before soldering? 

I did not take a photo before "liquid metaling" them and they fell after two years. 

My plan it is to do like this and use 3Momh directly (the pcie shunt is untouched) i did a photo:

 

https://ibb.co/jDCK21p

 

Thanks! 

Hi!

 

Maybe a little late... sorry.

Like @niobium615 Said, the orientation doesn't matter. I do want to suggest you pick a resistor with the same dimensions.
If you need further help let me know.