Timely maintenance will increase the life of your graphics card. First of all, this applies to graphics accelerators used in mining.
If you neglect to clean the cooling system of the video card, this can lead to fatal malfunctions. Possible options are chip or video memory dump, as these are the highest temperature-loaded elements.
How to understand that it’s time to take care of the video card? This can be identified by characteristic features:
If you neglect the timely maintenance of the video card, artifacts may appear, indicating that one of the microcircuits is in the dump. In this case, it may be necessary to replace the video card chip, and this is a completely different story.
If the video card was used in mining or a gaming computer, it becomes clear how often it is necessary to clean the cooling system.
It is desirable to service video cards:
Why is it recommended for miners to perform maintenance so often? This is explained by the fact that the video card works constantly (24 by 7) with an increased load on the memory and chip. At the same time, the VK thermal interface loses its adhesive properties and ceases to transfer heat to the radiator. Thus, the chip heats up more, because of which the solder begins to degrade and the chip breaks off. That is why it is desirable to service the video card during mining, at least once a year.
Remember that all the work you perform on the disassembly and maintenance of the video card, you perform at your own risk. Accuracy is the main quality of any repairman. Hands are recommended to be degreased periodically, especially when cleaning the board. If your service center has a ground loop, you should use an antistatic wrist strap.
Visual inspection is an integral part of any repair. What to look for when inspecting a video card:
Video card disassembly algorithm. Using a Phillips screwdriver, unscrew all visible screws on the back of the coolers. On some VK models, the screws may be located under the serial number sticker. In order to save the sticker, with a hairdryer set to 160 degrees, we warm up the location of the sticker. As a rule, the screws that secure the cooling radiator with a spring. It is important not to confuse them during assembly.
If there is a Backplate, unscrew all the screws that secure it to the board. When servicing a video card, it is necessary to remove the backplate, since under it there may be thermal rubber bands designed to cool DrMos.
If the card has not been serviced for a long time or worked at high temperatures, without proper cooling, the thermal paste dries up and starts to break down, the heatsink can “stick” to the crystal. In order not to damage the chip, when disconnecting the heatsink, it is necessary to perform small rotational movements. In no case:
After removing the cooling system, carefully inspect the location of the connectors. Disable cooler and backlight cables (if provided in the design).
In some designs, a separate heat sink for cooling the power phases is provided. And the VRM gives off heat through the Velcro, which needs to be replaced.
Gently, with a dry cloth, clean the GPU crystal from the old thermal paste. If the heat-conducting paste gets between the capacitors on the chip, it can be removed with a wooden toothpick, while being extremely careful. If you accidentally “demolish” the capacitor, then it is installed in its original place, with the video processor soldered off the board.
We inspect all thermal rubber bands, if they are dry and (or) damaged, they must be replaced.
Clean the motherboard from grease escaping from the thermal pads. You can use a toothbrush for this. Performing circular movements, with almost no pressure.
During maintenance of the video card, the toothbrush must be periodically washed in a degreaser so as not to smear dirt all over the board.
The cooling radiator must be purged with a compressor or blower. It is not recommended to use compressed air for this purpose. Since with a long press on the spray button, the temperature of the gas coming out of the can drops until moisture appears. This is the so-called Joule-Kelvin effect.
Thus, when using Duster or its analogues, condensation may form on the video card board. which needs to be dried for a long time.
Most often, thermal pads on memory chips lose their properties during mining.
When servicing VC, it is better to replace absolutely everything with thermal pads. What to look for when choosing thermal rubber?
The first is thickness. You can not install thicker ones, because. in this case, the thermal pad itself will not transfer heat to the radiator, but will take the excess heat onto itself. It is also impossible to install thermal pads less than the thickness recommended by the manufacturer, in this case heat will not be transferred, because. there is no connection to the radiator.
Thermal pads are selected in thickness, comparing with previously installed ones. On video cards, the thickness of the thermal rubber bands installed on the memory and on the VRM transistors may differ.
If the previously installed thermal pad is destroyed, so that it is not possible to determine its thickness by comparing it with the newly installed one; in this case, act according to the following algorithm:
After removing the old thermal paste, apply a new one in a small even layer. Spread the new thermal transfer paste evenly over the crystal with a plastic spatula or index finger.
When servicing the GPU, fans with broken blades must be replaced. It is not recommended to restore the impeller using glue, as this disturbs the balance and causes a beating and an uncharacteristic sound appears.
Replacement sequence:
After servicing the cooling system and replacing the entire thermal interface, it is necessary to install the radiator in place and immediately screw it on. The radiator is screwed onto the screws on the reverse side of the chip crosswise, without overtightening. To avoid chipping the crystal.
Then connect all loops. Install backplate.
We install the video card in a test stand or in our computer. We launch the pre-installed program for stress testing FurMark video cards. We test the card for 30 minutes and monitor the temperature. If the previously indicated signs of overheating do not appear, the service of the video card is considered completed.