The kit includes:
"Narrow" thermal pad for inductors: ~1.8mm x ~27mm x ~109mm
"Wide" thermal pad for rear-side of PCB: ~2mm x ~56mm x ~109mm
Thermal compound. Ours was 5W/mK, but the kits will vary based on supply.
You will require a Phillips screwdriver for this. No additional tools are required. We would recommend a paper towel and rubbing alcohol for cleaning thermal compound.
Our tutorial video is below, if preferred, but we've also written this in article form.
Step 1: Remove the Cooler & Faceplate
We're going to start by installing thermal pads to the front-side of the PCB, which will bridge contact between the baseplate and the heatsink. This isn't an ideal solution, but it's the one that EVGA is encouraging right now (and has instructed). The faults are simple: In an ideal scenario, you'd have a coldplate of some kind to contact the thermal pad for wider surface area conduction, then stem into fins for dissipation by the airflow. In this situation, we're adding a thermal interface to facilitate the transfer of heat from VRM components to the cooler's aluminum fins. This is still significantly better than the existing solution, which is to basically rely upon air as a thermal conductor (hardly even 0.3W/mK thermal conductivity at 25C). We end up with a hotbox, basically, and the baseplate and backplate may even do more harm than good.
The thermal pad mod adds that interface between the heatsink and the components, bridging thermal transfer and allowing significantly improved VRM temperatures. Take a look at some of our testing images from our previous article:
Ground yourself before continuing or work in an environment with low risk of static electricity build-up.
First, remove the four spring-tensioned screws in the back of the card. These are the screws that secure the heatsink and faceplate to the card. They do not secure the backplate. We are not yet removing the backplate. You should only have to remove four screws.
Gently flip the card and lift the heatsink and faceplate about half an inch from the PCB. You'll see that it's loose, but still being held in place by two cables. These are for the LEDs and the fan. Disconnect the cables with gentle force (you can use a flat-head to apply some pressure to the connector). Do not pull on the wires; this action may rip the wires from the housing.
Step 2: Apply the Thermal Pad to the Chokes / Inductors
With the heatsink removed, you now have access to the baseplate, the GPU, and the inductors. EVGA's recommendation is to place the "narrow" pad (the smallest of the two) atop the line of inductors. If you have purchased your own thermal pads and are cutting them to size, cut them to be about 1.8 x 27 x 109mm.
Remove the plastic cling that flanks the thermal pads provided by EVGA. Discard it. Ensure there is no plastic between the thermal pad and the inductors or the heatsink. It should be somewhat sticky/tacky on both sides.
Step 3: Clean the Thermal Paste
With this done, dab a paper towel with rubbing alcohol and clean the thermal paste off of the GPU and coldplate. Take your tube of new compound (provided by EVGA -- though any non-conductive thermal compound of reasonable conductivity would be fine) and apply some to the GPU surface. This is not like a CPU where thermal compound is more sensitive. There is no IHS between the silicon and the heatsink, so using a bit more than you might normally use on a CPU is acceptable. We would suggest about the size of a grain of rice in the center. The compound will spread on its own when tightening the cooler back to the card, which should be done with opposing corners to spread the paste equally.
Step 4: Tighten the Cooler to the Card
Speaking of, reconnect the LED and the fan header, then line the card up and flip it carefully. Line the four holes in the PCB up to the backplate/baseplate. Insert one screw and tighten it about half-way (we do 3 turns), then the opposing corner, then the other two. Tighten them fully after all four are in half-way. This will spread the compound evenly. Anyone who tells you that this is "too much," "too little," or "uneven" distribution of compound doesn't know what they are talking about, and is buying into the idea that there is a singular, perfect method of application. The fact of the matter is that this method works well, and an alternative is not necessary. We have done this dozens of times.
Step 5: Remove the Backplate
There are thirteen screws connecting the backplate to the PCB. Each of these is surrounded by a metal circle to indicate its use for securing the backplate. You do not need to remove the heatsink's four spring-tensioned screws for this step, nor the I/O shield screws.
Once the eleven perimeter screws and two screws near the GPU's back-side are removed, you may pull the backplate away from the PCB. There will be four thermal pads near the GPU, but should likely remain stuck to the backplate. Leave these untouched.
The area between the four screws we've outlined is where we need the wide thermal pad:
FTW cards have three rubber bumpers that will need to be removed and discarded. These are only to prevent the PCB from contacting the backplate (shorting), but we actually want that contact now -- so get rid of any rubber bumpers in the way of the application zone outlined above.
Using the above as a guide, apply your new thermal pad using the same steps as for the narrow pad. Remove all plastic, drop the pad into place, and ensure no air pockets.
Place the backplate onto the PCB and re-secure it with the thirteen screws. You are now done.
If any issues or odd behaviors are observed while gaming, contact EVGA for additional support.
Editorial: Steve "Lelldorianx" Burke
Video: Andrew "ColossalCake" Coleman