Let’s see why in recent years video cards often fail due to the dump of the graphics processor or memory chips and how to avoid this. And we will also touch upon an important question for many – what wears out a video card more: mining or games?
Surely you have seen on the Internet photographs of color “artifacts” on the monitor screen, they appear if a video card has a graphics processor or memory chip. “Artifacts” are not the only sign. Dumping chips can be accompanied by a black screen when turning on the PC, inability to install drivers on the video card and error 43 in Windows, indicating system failures related to the graphics adapter.
Problems most often appear with video cards released in the late 2000s and later, and the Nvidia GeForce series video cards of the 8X00, 2XX, 4XX and 5XX families suffered most from dumps. Many of these models have not survived to our times in working order, they are often sold on the secondary market after homemade repairs using the warm-up method, but it does not help for long.
GeForce 8800 GTX was the record holder in chip dumps
What is BGA and why does chip dump occur?
To understand what a chip dump is and for what reasons it occurs, you first need to understand how to attach the microcircuits to the video card textolite . In the 1990s, video chips had very few pins, for them the DIP package (from the English dual in-line package), the pins of which are located along the edges of the microcircuit, or QFP packages (from the English Quad Flat Package), where the pins were from four sides.
Even such complex video chips of the 1990s as 3dfx Voodoo II had enough QFP cases
In the 2000s, the complexity of graphics processors and their power consumption began to grow rapidly, the number of pins reached hundreds of pieces or more, which made it impossible to execute them in QFP packages. The way out was the BGA (Ball grid array) type of package, in which contact is provided by solder balls located on the reverse side of the microcircuit. The number of pins also increased for video memory chips, which, after a short existence in TSOP (Thin Small-Outline Package) packages, also moved to BGA packages.
The video chip GeForce 4 Ti 4200 already uses the BGA package, while the video memory is still managed by the TSOP package.
BGA solved the problem of miniaturizing high pin count chips, but unlike DIP, QFP, and TSOP packages, the chip’s solder ball pins are not flexible. With repeated strong heating, followed by sharp cooling, microcracks and oxides appear in them, gradually leading to the so-called “dump” of the chip, when one or several balls lose contact.
The contact of the balls can also be disrupted by physical impact on the textolite or the chip, for example, if the video card is inaccurately installed in the PC. And also when dismantling its cooling system , sagging in a slot under its own weight or with strong and prolonged vibration.
Because of this, BGA chip packages are considered unreliable and are rarely used in electronics industries that require trouble-free operation despite temperature fluctuations or vibration, such as military equipment or aircraft construction. The situation with the unreliability of BGA cases worsened in the second half of the 2000s, when manufacturers finally switched to the use of environmentally friendly lead-free solders in the manufacture of consumer electronics and video cards in particular.
Lead-free solders have a higher melting point and higher hardness of the resulting balls used as chip contacts. If the softer lead-containing solder provided some plasticity to the soldering, then with lead-free solders, the contacts of BGA chips became even more susceptible to both mechanical and thermal damage.
In those years, forums were actively discussed by users involved in electronics repair. They assumed that with the help of lead-free solders, manufacturers first of all solved the issue of the planned obsolescence of the devices. This is not surprising, because after the video cards and motherboards of the 1990s, working for 5-10 years, users began to face chip dumps after a couple of years of device operation.
Can the chip dump be repaired?
Dumps of chips gave a whole direction to the artisanal method of repair, which was called “heating” or “roasting”: users heated the video card in different ways – from an oven and an iron to a hair dryer. Usually such a “repair” helped, but for a very short time. After a couple of months, the user again encountered a chip blade and video card artifacts.
The fact is that the solder balls are located not only under the chip substrate, with which it is attached to the video card textolite, but also between the chip and the substrate, where their size is much smaller. And most often, the dump and breakdown of the contacts of the solder balls occurred precisely between the chip and the substrate.
Experienced computer repairmen can repair the blade between the textolite of the video card and the chip substrate by making the so-called “reball” – the solder balls are replaced with new ones using special stencils and subsequent soldering.
But the dump of balls between the substrate and the chip is practically irreparable. In this case, only transplanting a working chip, for example, from a “donor” video card, will help. The video memory chip blade is also repaired using a “reball” using a working chip.
What causes chips to dump during normal use of a video card?
Let’s imagine a typical use case for a gaming graphics card. When you turn on the PC, the graphics processor of the video card heats up from room 20-25 degrees to 35-45 degrees in idle mode. In the case of using the “FAN STOP” function, which nowadays is increasingly used by manufacturers, a video card in idle time can warm up to 50-60 degrees. It depends on the quality of its cooling system and the ventilation efficiency in the case.
“FAN STOP” not only reduces fan noise and wear, but also reduces the amount of dust on the video card
When you start a game, the temperature of the video processor usually rises to 65-85 degrees – again, depending on the cooling efficiency of the video card and the PC case. But the temperature constantly jumps when the load on the video card decreases, for example, when entering the inventory, loading a level, or playing a cut-scene.
That is, the BGA-chip solder balls are constantly exposed to temperature differences, causing them to expand and contract. And with frequent folding of the game into the tray and deploying it back, temperature drops can be up to 40-50 degrees, which even more hits the solder balls. In a few hours of play, there can be several hundred such heating and cooling cycles.
Why are video memory chips vulnerable to dumping?
The temperatures indicated above are normal for a video processor, we usually see them in monitoring programs such as MSI Afterburner or HWiNFO. But the temperature of video memory chips is often not monitored on video cards of the budget and middle segment, although their heating can reach much higher values than that of a video processor.
The situation is aggravated by the fact that manufacturers often do not pay due attention to cooling the video memory, focusing on cooling the video processor. As a result, we get a video card with a cold video processor that needs a small flow of air for cooling, but with video memory, which is not enough for proper cooling.
A striking example of such video cards is the ASUS GeForce GTX 1060 Strix with a massive and redundant CO on a video chip, and a small plate that should cool the video memory, but does not even cover all the chips.