DIY video card repair geforce 9500 gt

Have you ever experienced your graphics card stopped working? God grant, of course, that this never happens, but still! What to do, for example, if you hear that the computer is booting, but there is no image on the monitor (black screen)?

What we usually do in such cases: substitute a known working video card (or switch to integrated video) and make sure that the problems are with the graphics adapter. But what to do in this case? Can we provide video card repair ourselves?

The good news is yes: DIY video card repair is quite possible! The bad thing is that after such a repair there is no guarantee that the video card restored in this way will work for a long time. Also, the repair itself may end unsuccessfully if we do not follow certain rules. But let's talk about everything in order! 🙂

So, we have a non-working video card from Nvidia, the GeForce 9500 GT model. Like this:

What's the problem? The video card worked for a long time in harsh temperature conditions, which led to its overheating. As a result of this, a fairly typical (in such cases) thing happened: the "dump" of the BGA chip of the video card.

Don't be scared by the word “dump”, nothing has fallen off there 🙂 This is just how the people call the electrical contact of the BGA balls array with the printed circuit board of the card that occurs as a result of prolonged overheating. Usually, this phenomenon occurs as a result of the presence of a small section of cold soldering, which is subjected to prolonged and strong heating.

This is not to say that this is a 100% marriage of the manufacturer: there can be quite a lot of tin balls in the array and the violation (or oxidation) of the contact of even one of them can lead to a complete (or partial) loss of the card's performance. So overheating, be it a video card or a central processor, is a very unpleasant thing. Try to avoid it in every possible way!

And in this situation, we have no choice but to try to repair the video card with our own hands, on our own. So, first of all, we need to take care to remove all plastic plugs on the card, stickers (stickers) located on the back side from the card. Anything in the area of ​​the graphics chip can melt.

Yes Yes! You heard right: to melt. After all, we will repair the video card by heating it and all the “unnecessary” must be removed, just for every fireman. Of course, perhaps nothing like this will happen, but just make it a habit - it will come in handy 🙂

We will also need to remove the fan and cooling system. We do this with whatever is more convenient for us. Personally, I use this screwdriver with interchangeable nozzles:

We unscrew the screws securing the fan, remove the metal cover and get the following picture:

As you can see, the cooling system requires thorough cleaning, and the fan itself needs preventive maintenance, since its efficiency has decreased due to lime adhered to the blades and dust crammed into the bearing.

The next step is to remove the GPU heatsink. It would seem: what is so difficult here? But, as one movie hero said in a film about various evil spirits: "There is a catch everywhere!" Here it lies in the fact that often (especially if the chip was operated in a severe temperature regime), when dry, the thermal paste tightly glues the crystal and the heatsink together.

It is categorically not recommended, in this case, to use heroic force and pull this matter upon yourself, or, as experience erroneously suggests, to pick something up! This can damage the crystal! There is a simpler and more elegant solution: we take a regular household hair dryer and slowly warm up the clutch area.

After a while (5-10 seconds), we begin to slightly shake the radiator from side to side, as in the photo above. Thermal paste, softening under the influence of temperature, will allow us to do this. Warming up this thing a little more, we can easily separate our radiator from the crystal:

Try as completely and as carefully as possible to clean both the "base" of the radiator and the crystal itself from the remnants of the old dried thermal paste. At the same time, try not to scratch the metal surface of the radiator (this will reduce its heat transfer coefficient). Do not scrape, it is better to warm up separately and wipe off the old paste.

With the crystal it is also as careful as possible: if it was not possible to remove some part of the paste (like mine, for example), then it is better to leave it. Anything that is erased - be sure to delete! Otherwise, the thermal paste, under the influence of temperature, will, as they say, “bake” and then it will be very difficult to remove it without damaging (chipping) the core itself.

Before we start repairing a video card with our own hands, let's take a closer look at the graphics chip.

Why did I highlight some of the areas in the photo above? See, the larger area is the graphics card chip itself, and the smaller area is the GPU die (GPU - graphics processing unit). Along the perimeter of the crystal, we see a white sealant (compound) that performs several functions: protects the crystal from dust getting under it and fixes it to the substrate.

What is the "trick" here and why repairing a video card with your own hands can end unsuccessfully, no matter what efforts we make? The area (array) of BGA solder balls is not only between the chip itself and the PCB textolite, but also between the crystal and the substrate graphics card!

Do you feel where I'm going? The harsh reality is that we can repair the video card ourselves (if we are still lucky) only if there is a violation of the contacts of the balls directly between the PCB and the substrate. If the “dump” occurred under the crystal, then we can hardly do anything about it. Even such an operation as reballing (complete replacement of an array of balls using a stencil) will not save in this case, since this procedure is performed only for the “bottom” of the entire chip, but not for the crystal!

So, I hope we have learned the necessary minimum of theory? Moving on! To repair a video card at home, we need a flux and a disposable syringe. I use the usual GFR (alcohol-canin), which is called “GFR-flux”.

We collect the substance in a syringe (about one cube). If it remains, it will be possible to drain back.

Note: You can use any other inactive (ideally neutral) flux. For example, "F1" or "F3". The original LTI-120 is also suitable. Although, with LTI, not everything is so simple: leave it as a last resort 🙂

We carefully apply the tip of the needle to the edge of the substrate, tilting it so that the flux we squeeze out of the syringe is under the chip. After pumping it, if necessary, tilt the card slightly so that it spreads well between the balls. Ideally, we want to achieve an effect where the liquid appears a little from all sides.

Advice: after use, rinse the syringe (just take several times of water from the tap and squeeze out through the needle). If this is not done, the rosin in the needle will dry out and clog it. It will take a long time to clean or throw away.

Now we can be sure that when heated, the flux will fulfill its function. Why fluxes are needed, what they are and how to use them correctly, we considered in a separate article, so we will not repeat ourselves.

After that, we can proceed directly to the repair of the video card with our own hands! To do this, we place it in such a way that we have free access to the GPU from above and below and, using a soldering station, we begin to heat the substrate around the perimeter.

Note: under no circumstances heat the crystal itself! It can fail!

How I do it, I’d better show you in video format, since you cannot clearly illustrate with photographs alone.

Now let's comment on this video a little.When you warm up the video card from below (under the chip), try to keep the hair dryer perpendicular to the plane of the PCB, otherwise I didn't manage to shoot and warm it at the same time. Also, be careful not to flare the small card components located on the back side (they can be easily shifted, given the heated solder underneath).

In the video above, I did not show the whole procedure, as you understand. The bottom must be heated long enough (3-5 minutes) so that the smoke from the flux, which you may have noticed, began to rise quite intensively above the board (this is evidence that the board has warmed up well). The first stage will be “boiling” and bubbling of the flux - this is normal.

Also, do not hesitate to warm up the place under the crystal itself (you can do this through the board). The main thing: do not hold the hair dryer in one place - smoothly move it over the area (to exclude places of local overheating of the surface). Keep the funnel of the hair dryer at a distance of 2-3 centimeters from the surface to be treated. I personally set the air flow to an average value, the temperature that the soldering station shows at the same time is 420-450 degrees Celsius. The second value is the limit for my "Ya Xun 880D".

The temperature spread here is due to the fact that its sensor itself is located directly in the handle of the hot air gun, and the air temperature at the outlet of the hot air gun is already different (lower). Plus, here you can add the inevitable heat loss due to the ability to absorb and dissipate heat by the treated surface itself, the temperature in the room, the proximity of the hair dryer to the heated area, the power of the air flow, etc. That is why only empirically it will be possible to select the exact value of the operating temperature (thermal profile) for a particular soldering station.

To what state should you warm up? Here, again, there are indirect signs by which we can navigate. The whole procedure takes approximately 5-8 minutes. The spread of time is due to the factors listed above. It also depends on the quality of the flux used, the type of solder from which the BGA array is made on a substrate (lead or lead-free). In the process of strong heating, the flux should evaporate (smoke) fairly decently.

Another important marker is the visual detection of the melting of the solder on the elements located on the chip around the die (usually a series of small SMD capacitors). When the solder "shines" on them, it is a sure sign that the balls of the substrate have reached their melting point, which is exactly what we need! For greater confidence, you can take tweezers in your free hand and try to slightly move the chip itself: gently push it (literally by a millimeter) to the side and you will see how it “wobbles” and, due to the forces of surface tension of the molten balls from below, will snap into place. After that, the heating can be safely stopped!

Note: some craftsmen, instead of a station, use a regular building hair dryer or repair the video card with their own hands, “baking” it in a household oven, after wrapping it in foil! Frankly, I am not a fan of such radical methods of "repair", although (if the guys succeed in everything), then why not? 🙂

During the heating procedure, you can monitor the surface temperature with a thermocouple or pyrometer (infrared thermometer). This will help in the future to better navigate in the selection of the correct thermoprofile.

Note: when cooling the video card (and any other element), do not use forced airflow - fan, etc. Let the part cool naturally, no need to "fit" it. We don't need the microcircuit to receive a thermal shock (shock), do we?

This is how you repair a video card with your own hands! Whether they were successful or not, we have yet to verify. To do this, we need to do a few mandatory things. Out of habit, I clean (where possible) the board from the flux residues. In this case, it is the rosin remaining after the evaporation of the alcoholic component. The rosin is neutral (it does not interact with the components of the board) and, in theory, it may not be necessary to wash it off, but, for the sake of order, let's go over it thoroughly with a brush and a cleaner.

We washed more or less (the rosin dissolved), let it dry and apply fresh thermal paste to the crystal (“KPT”, “AlSil” or “Zalman” - I respect):

Now we put the whole "constructor" back together (we attach the heatsink, fasten the cooler, connect it to the connector on the board).

Before installing the card into the system unit, let's go (just in case) with an elastic band over the contacts of the Pci Express connector and that's it - can you install a component on the motherboard to check what we've got?

But it turned out, as we can see, everything is not even bad. There is an image on the monitor! Do-it-yourself video card repair is possible! Of course, to be completely sure, we need to install the operating system (it was not at hand), install the video card driver and, ideally, run some kind of stress test for stability, which will show us, in the end, we managed to repair the video card ourselves or No?

Note: The free and easy-to-use tool "FurMark" can work very well for this test.

Otherwise anything can happen: the video card seems to work, but the driver is not installed or does not pass the stability test. Also, as you understand, we cannot give any guarantee for this type of "repair" and we do not know how long the device will work? But, as they say, on the other hand, we “pumped over the skill” of repairing video cards at home, and the client received a temporarily working computer. We did what we could, and then there will be what should happen!

As always, I am waiting for your comments, feedback, constructive criticism below under the article 🙂

Static is when a high-voltage arc breaks through about 1000 volts and above (this is not specific information

). There was simply no common grounding of the two devices. Both have switching power supplies, in which there is a capacitor between the "hot" and normal mass. It is through it that an alternating voltage of up to 115 volts enters the usual mass (I personally checked it). So it turns out that there is a 115 volt difference between the devices and when they are connected together, the voltages are equalized. If you first touched the corus tires, then nothing terrible would have happened, but in this case, the voltage equalization passed through the signal-chipset-mass chain. This is how the chipset burned out.

PS: There was a case when I was waiting for some body to be sewn, then fidgeted on the chair and decided to stretch my legs (it turned out in the direction of the battery). and in his hands he held something, connected via a yusb cord. Well, he touched on the radiator to the place where there was no paint - then I got so fucked up through my leg-body-arm that I flew with a chair to the other end of the office (+ the surprise worked). How to spread your legs

Rebooting the video chip is useless - it either works or it doesn't. There is no third. If the artifacts are gone, this is death throes. You can warm up a video chip crystal at a temperature no higher than 360 degrees and no more than 5 minutes. After that, it will live for some more time. The video chip will withstand such warm-ups 2-3 and then die completely.
Rebol them only if they need to be supplied from the donor, or if after some water the laptop got caught, and in most cases either warming up or replacing with a new one.

Unfortunately, these are not telephones. I would have become a millionaire if reball helped. why skate there with such a step?

Believe it or not, the 440GTS threw a lot at the battery, removed the radiator from the core, fired directly into the radiator with a piezo-snag from a lighter (well, you know what I mean), repeated 40-50 times, then “gifted” the memory with such experiments, the result is how it worked and remained working ("adjusted" with a sidekick to the guarantee).

Such a problem I start the game (Doom 3, Two worlds, etc.) and after a short time the screen goes blank, i.e. the video card is cut down and you have to reboot. I tried to change the driver, DirecX 11 (maybe because of it).

CPU Type QuadCore Intel Core 2 Quad Q9400, 2666 MHz (8 x 333)
Gigabyte GA-P31-S3G motherboard (3 PCI, 3 PCI-E x1, 1 PCI-E x16, 2 DDR2 DIMM, Audio, Gigabit LAN)
Intel Bearlake P31 motherboard chipset
System Memory 2048 MB (DDR2-800 DDR2 SDRAM)
DIMM1: Samsung M3 78T2863RZS-CF7 1GB DDR2-800 DDR2 SDRAM (6-6-6-18 @ 400 MHz) (5-5-5-15 @ 333 MHz) (4-4-4-12 @ 266 MHz)
DIMM3: Samsung M3 78T2863RZS-CF7 1GB DDR2-800 DDR2 SDRAM (6-6-6-18 @ 400 MHz) (5-5-5-15 @ 333 MHz)
Video adapter NVIDIA GeForce 9500 GT (512 MB)
Monitor Acer AL1916W [19 ″ LCD] (6250C9B14010

Temperature (Everest): the temperature of the GPU diode is 68 (in games it rises sharply to 80-86, and maybe it did not measure further), the temperature of the CPU is 42, 32, 42, 40 for each of the four cores (when playing, it rises by a couple degrees).

The computer is new (I bought it at the beginning of last year) and it all started recently, it was not observed before.

The site, Do-it-yourself Computer Repair offers a set of practical information on repairing, setting up computers and Windows with your own hands. Here you can find tips on how to fix this or that computer malfunction on your own. How to install and configure Windows. In addition, the site regularly publishes interesting news and articles.

This is where the outputs of the video card are located. Please note that the slot panel of almost every expansion card is accessible from the outside of the PC case. Therefore, all the necessary inputs and outputs are located on it.

A computer is a digital machine, so the digital format for a computer is "native", it is better to use it to connect a monitor to a video card. Modern displays have come a long way from the first cathode ray tubes (CRTs) to liquid crystal displays (LCDs). CRT monitors are analog in nature, so for them the digital signal is converted to analog using a digital-to-analog converter (DAC), which is located on the video card. With the advent of liquid crystal displays, the need for a DAC has disappeared, but this component is still present in case of connecting analog CRT monitors.

Composite video output “cinch”, also known as RCA (Radio Corporation of America) connector.

The component outputs are too large to fit on a graphics card, so an adapter is almost always used. Usually the adapter provides component video (the first three connectors) and sound (the last two connectors). This standard provides three separate cinch connectors: “Y”, “Pb” and “Pr”. They provide separate color information for HDTV (High Definition Television). This type of connection is also found on many digital projectors. Although the signal is transmitted in analog form, its quality is quite comparable to that of the high-definition VGA interface. High definition (HD) video can be transmitted via the component interface.

HDMI stands for “High Definition Multimedia Interface”. HDMI - future standard full description of HDMI. This is the only interface that provides the transmission of video and audio information over a single cable. HDMI was developed for television and film, but computer users will be able to rely on HDMI to watch high definition video.

Graphics card interfaces
With its interface part, the video card is inserted into the motherboard of your computer. In fact, this is a slot through which a computer and a video card exchange information. Since a motherboard usually has a slot of one type, it is important to buy a video card that will fit it. For example, a PCI Express graphics card will not work in an AGP slot.

The PCI interface is a modern standard for most expansion cards, but video cards at one time moved away from the PCI interface to the AGP standard (and later to PCI Express). Some computers do not have AGP or PCI Express slots for graphics upgrades. The only option for them is the PCI interface, but video cards are rare for it, are expensive, and their performance leaves much to be desired.

PCI-X stands for “Peripheral Component Interconnect - Extended”, that is, we have a 64-bit bus with a bandwidth of up to 4266 MB / s, depending on the frequency. PCI-X (not to be confused with PCI Express!) Is the first high-speed upgrade of the PCI Express bus, but at the same time it has received a number of functions useful in the server space. The PCI-X bus is not very common in regular PCs, and PCI-X graphics cards are very rare.You can install a PCI-X card in a regular PCI slot if it supports the latest version of the standard (PCI 2.2 or higher), but is not PCI-X compatible with the PCI Express standard.

AGP - high bandwidth interface specially designed for graphics cards. It is based on PCI specification version 2.1. The AGP interface went through several versions, and the latest was AGP 8x at 2.1 GB / s, which is eight times faster than the initial AGP standard at 266 MB / s (32 bits, 66 MHz). AGP on new motherboards gives way to PCI Express, but AGP 8x (and even AGP 4x) still provide sufficient bandwidth for modern video cards. All AGP 8x cards can work in both AGP 4x and AGP 8x slots.

Unlike ISA, PCI, and AGP, PCI Express is serial, not parallel. Therefore, the number of contacts has decreased significantly. Unlike parallel buses, the required bandwidth is available for each device. Whereas, for example, for PCI, the bandwidth is divided between the cards used.
PCI Express allows you to combine multiple single lines to increase bandwidth. PCI Express x1 slots are short and small, yet they provide a combined speed of 250 MB / s in both directions (to and from the device). PCI Express x16 (16 lanes) provides bandwidth of 4 GB / s in one direction or 8 GB / s in total. Smaller PCI Express slot options (x8, x4, x1) are not used for graphics. It should be noted that mechanically the slot can correspond to x16 lines, but logically fewer of them can be connected to it. There are many motherboards that have two PCI Express x16 slots that can run in x8 mode, which allows you to install two video cards (SLI or CrossFire).

Graphics cards can consume (and therefore release) as much power as a 150-watt light bulb. This amount of heat generated from the surface of a single silicon chip can easily burn the die. Therefore, heat should be dissipated in a timely manner using stable and powerful coolers. Without cooling systems, the graphics processor or memory can overheat, leading to a "freeze" of the computer, and in the worst case, even to the failure of the video card.
Cooling can be carried out both passively with the help of heat-conducting materials and radiators, and actively, if the fan is running. But in the latter case, you will have to be content with an increased noise level.

Heatsink is usually understood as passive cooling. The heatsink lowers the temperature of the chip to which it is connected by removing heat and increasing the heat exchange area with air. For this purpose, radiators usually use fins. They can be found on GPUs as well as memory chips.

Here I described my problem -> click

I decided to buy a new video card, namely the model that was originally (9700). Here, I found something. Is this normal or can you recommend another product (of the same model)?

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

brsgvrn,
Can you give a command for “NVFLASH”? To flash 9500M GS to 8600M GT.

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

In addition, I can suggest the following. Look carefully under the BIOS sticker (U3) which microcircuit is there. If the company is ATMEL, and the numbers of the video cards (sticker above the chip) start with 7A. 7B, these microcircuits are not flashed. Only their replacement with other firms. Personally, I sewed with a programmer.

Comment: This post has been moved from the Upgrade and Upgrade Laptop topic. Moved: reylby

In the summer of 2008, Nvidia announced its next creation for gamers, releasing the GeForce ninth series video card, while positioning it as a solution for inexpensive home or office computers. The video card was named Nvidia GeForce 9500 GT. Technically, the graphics card comes in several versions: with 256, 512 and 1024 MB video memory.

The architectural closeness of the adapter to the previous model Nvidia GeForce 8500 GT is explained by the same model of the GPU, but with some improved parameters, for example, an increased clock frequency up to 550 MHz and a 55 nm process technology.

The interface for connecting to the motherboard, as in the previous model, is made via PCI-E 16x version 2.0. The maximum resolution of the video card is 2560x1600 pixels, while the physical connection of two monitors at the same time is supported. RAMDAC frequency is within 400 MHz.

The clock speed of the graphics processing unit (GPU) is at 550 MHz, and the frequency of the shader units is 1400 MHz. The type and amount of video memory for each individual manufacturer differs in the following options:

• 256 and 512 MB with GDDR3 memory with 128-bit memory bus;
• 1024 MB (from the developer GIGABYTE) with GDDR2 memory type and the same 128-bit memory bus.

The video memory frequency in the version with GDDR2 is estimated at 1000 MHz (which is exactly 200 MHz higher than the indicators of the previous 8500 GT), and in the GDDR3 version - 1400 MHz.

The specifications of the Nvidia GeForce 9500 GT are as follows:

• number of universal video card processors: 32;
• number of TMUs (texture units): 16 with support for shaders version 4.0, which adds even more realism to the gameplay;
• the ROP number is 8;
• maximum FSA degree: 16x;
• supported standards DirectX 0 version (outdated standard), Open GL 3.2, although modern video cards use versions 4 and higher;
• supported technologies NVIDIA PhysX, CUDA, Direct Compute0.

Another obvious advantage over other video card models is support for the SLI / CrossFire connection mode - the so-called Multi GPU. This means that the user will be able to combine the power of two identical video cards by connecting them through a special MIO connector in order to achieve the highest performance several times over in all respects.

At the same time, the manufacturer notes the need for a computer power supply of at least 550 W in order to maintain the required power supply.

The printed circuit board with original inscriptions and without an additional power connector is made in a compact style similar to the 8500 video card. Under the cooler there is a monolithic cooling radiator providing high thermal conductivity.

The normal temperature of the GeForce 9500 GT video card, observed at low graphics and multimedia loads, is 45-60 degrees. It can go up to 70 or more under high GPU loads.

Overclocking means an increase in some indicators of a video card to the maximum in order to provide the highest performance, albeit taking into account the increased power consumption and high heating of the GPU.

To do this, it is recommended to use the free software Nvidia Inspector or its analogue MSI AfterBurner, which step by step increase the processor frequency (GPU Clock value), memory frequency (Memory Clock) and frequency of shader units (Shader Clock).

It was found that overclocking Nvidia GeForce 9500 GT with DDR2 memory and 512 MB RAM can be done up to the following values:

• GPU Clock: 620 MHz;
• Memory Clock: 600 MHz;
• Shader Clock: 1500 MHz.

The basic rule when overclocking any video card is gradualness. Each individual value should be increased by several MHz and the results should be tested in the game. If the allowed values ​​are exceeded, the adapter will reset the values ​​to the standard values, and the computer may freeze, requiring a system restart, or artifacts will appear on the screen.

If, according to the standard parameters, some games cannot produce the desired graphics settings, then overclocking the Nvidia GeForce 9500 GT video card will help increase the card's potential by several percent, but not more than 40%. It is also necessary to make sure that the cooling cooler is working correctly and that there is thermal paste in order not to damage the video card.

Let's take a look at overclocking a GeForce 9500 GT using the Nvidia Inspector program (version 1.9) as an example:

1. We launch the program.
2. At the bottom of the window, click on the "Show Overclocking" button. The program will display a confirmation request, we agree - click "Yes".
3. The Nvidia Inspector window will be expanded with additional settings and up-to-date information about the processor frequency, memory frequency and shader frequency.
4. By moving the slider or using the (+1, +10, +20) buttons, we increase the GPU Clock value from 550 MHz by several points, for example, to 580 MHz.
5. We perform the same action with the Memory Clock values ​​(the maximum can be increased by 50%, but no more) and Shader Clock.
6. Click on the "Apply Clocks & Voltage" button to apply the changes. Then we check the performance results in the game.

It should be borne in mind that when you turn off or restart the computer, the values ​​set through the overclocking program are reset to the default values ​​set by the developer. To avoid this, in the Nvidia Inspector window, click "Create Clocks Dhortcut". After that, the program will create a shortcut on the desktop, by clicking which it will automatically set the saved overclocking values.

Some users use a video card to earn cryptographic currency on the Internet, although it is recommended to use more modern versions of graphic cards that support high bandwidth for this.

Relatively ancient by today's standards, the 9500 GT can be used for mining through NiceHash, which is less demanding on the performance of a video card. If we compare the electricity consumption and the income received, which allows at least to pay for the use of the services of an Internet provider, then mining on 9500 GT is not advisable. For this, it is better to use a new type of video card.

Relatively non-demanding games released before 2008-2010 run at medium and even high rates and maintain good FPS values ​​(FPS is the number of frames on the screen per second).

Even more demanding toys released later will start, but they will not be able to play calmly. In addition, the video card does not support DirectX 12 and Open GL 4.

The test in the games GeForce 9500 GT showed good results, which are indicated in the table below. The first column contains the names of the games, the second and subsequent ones - the number of frames per second (FPS) in the screen resolution specified in the column. The higher the FPS, the more stable the gameplay.

DIY modding - modding alexgo

In this post I want to talk about my experience roasting (warming up) video cards, as well as unsuccessful examples of roasting video cards by other people. I repeat once again that everything is based on personal experience. Therefore, my assumptions and conclusions may be wrong. Nevertheless, almost all the video cards that I have warmed up work to this day.

This entry is not a guide to action. I, as an author, do not bear any responsibility for the damage to your "hardware" by your own stupidity or as a result of crooked hands.

So why do video cards warm up? Video cards warm up when a video chip or memory chips have fallen off. It fell off not in the sense that it fell off the board (from the PCB), but in the fact that contact disappeared at some points between the chip and the board.

Why is this happening? Here I cannot give an exact answer, but I can only assume that chips or memory (much less often) fall off the board as a result of prolonged exposure to high temperatures. And also, possibly, due to the relatively sharp temperature changes "hot-cold-hot". One way or another, the connection between the contact on the chip and the contact on the board disappears. The solder balls remain either on the chip or on the board. Multiple contacts not working and causing video cards to malfunction.

As a rule, this is found in "hot" (ie, with high heat dissipation) video cards and usually appears after a year or two of using the video card. For example, the well-known series of video cards GeForce 8800GTSGTX (both on the first "hot" G80 chip, and on the subsequent "colder" ones). But in my practice there were also colder video cards, for example, nVidia GeForce 6600GT, GeForce 8600GT, Radeon x1600Pro.

What are the symptoms of the chip "dumping" (loss of contact)? The symptoms are as follows: artifacts - vertical stripes of color on the screen, colored porridge (a mess, as some call it). Also, several times the following was encountered: without the installed drivers, the picture was displayed normally, with the installation of the drivers, either there were artifacts, or the system went into BSOD ("blue screen of death").

(Sample photos were taken in this entry .)

How to treat such an ailment? Here, in fact, we come to the most important thing - roasting a video card.

I will first talk about unsuccessful other people's examples frying video cards. So the most common way to warm up video cards is roasting in the oven... Yes, in the oven of an ordinary gas (or electric) stove. This method is very popular, probably thanks to the resource from his Personal pages , where you can find examples of warming up video cards. It is not for nothing that I singled out this method in unsuccessful examples. I find roasting graphics cards in the oven a bad way.

Why bad, you ask. Because in this case, the entire video card warms up, i.e. the impact goes to all components of the video card. Which I consider not acceptable. The simplest thing that can happen is the elements made of plastic melt. For example video connectors DVI, VGA and others, as well as power connectors. Naive users believe that if they wrap these elements in foil, it will save them from melting. As practice shows, this is not the case. Whether they are wrapped in foil or not, they will melt, since the video card will warm up evenly in the oven, the foil will only slow down this process. But slightly melted connectors are not a problem. Problems can arise when other elements of the video card overheat - various electronic components whose maximum permissible temperature is much lower than the temperature of the chip warming up. They will simply fail and we will finally add the video card.

Based on these facts, we can make some conclusions... Namely:

• Warming up of the video card must be spot-on, i.e. only the chip and memory microcircuits should warm up, and not the entire board as a whole.
• The temperature, in my experience, should not exceed 220 degrees Celsius to warm up the chip.
• Warming up must be carried out evenly, it is highly desirable with a gradual increase in the heating temperature.
• The warm-up time should not exceed 5 minutes for a relatively fast heating and 10 minutes for a smoother one.
• The video card should cool down smoothly, there is no need to blow over it additionally.
• During and after frying, it is highly advisable not to displace, move, or shake the video card. Otherwise, as a result of these actions, small components located near the chip may move or fall off, as the solder is melted.

Now about my experience in frying video cards. When I first decided to "repair" video cards in this way, I immediately gave up roasting in the oven. And in the absence of special tools, he did it over the burner (ceramic and electric tiles are also used), i.e. over an open fire.

This method has both pluses and huge minuses. The pros are that we can warm up the video card pointwise, smoothly heating and cooling. Cons - open fire, the possibility of darkening the video card or burning paper stickers (stickers) glued to the video card. Nevertheless, I worked out a certain warm-up strategy for myself - the video card was 20-25 cm higher than the fire, did not stay in place for a long time (it moved it smoothly, evenly warming up the chip and memory). In this way, many video cards were cured. 😉 For example, in my computer the free GF 8800GTS 640Mb still “lives” and works fine (for about half a year) in overclocking mode.

My next step towards improving the quality, reliability and safety of repairing video cards by the frying method was the use of a hair dryer... For example, I use a Bosch hair dryer. It is convenient in that it has digital control, has several operating modes and allows you to set the required temperature, as well as the power (speed) of the air flow.Working with a building hairdryer, for repairs, it takes me about three minutes to warm up itself and a few minutes to smoothly natural (without additional blowing) cooling of the video card.

Using a hair dryer is the most optimal way to fry video cards.

You can arrange holivars on the topic of whether it is worth warming up video cards or not. When contacting a service center with similar symptoms for a video card, you will probably be advised to throw it out. I know that the SC masters are categorically against such a "repair". In fact, they are right that having repaired the video card in this way, they will not be able to give any guarantees for the duration of the work.

By the way, about the duration of the video card after frying. Depending on whom at what temperature the video card was warmed up, how long it was warmed up, and the normal operation time of the video card will differ. It can range from several days, weeks, months, up to several years (at best).

Now let's look at a few examples of frying video cards, both successful and unsuccessful.

A good example. My video card is GF 8800GTS 640Mb. In the photo before frying. Unfortunately, there are no photos after roasting, or rather there are, but only the sides with the chip. After frying, nothing melted over the burner, only the stickers that could not be peeled off darkened. The video card had to be fried several times. The last time it was a long time ago, I can no longer remember. Now the video card is cooled by the SVO, so there are no problems with overheating.

On the Personal pages of Overclockers.ru you can find several examples of roasting such video cards. And almost all of them are made with an oven.

Here's another example roasting the XFX GTX 280 graphics card in the oven ... In principle, the video card worked, but some of the plastic elements were melted.

And then completely unsuccessful example of frying a video card in the oven. Subject Palit 8600gt 256mb sonic +.

As you can see, the person "slightly" overdid it. As a result, he killed the video card. Not only the electrolytic capacitors, but also the solid-state capacitors (which can withstand high temperatures) swell from the video card. The protective frame on the chip around the crystal, DVI connectors, and the fan power connector are also melted.

Not long ago, one person also ditched the excellent 9800GTX + video card. But I told him not to heat it up in the oven. Eh, sorry for the vidyushku. Unfortunately there is no photo. Perhaps it will be.

So, everyone decides for himself whether to warm up the video card or not. If you think it's worth it, then I advise you to contact a person with experience who has the right tool in stock.

In addition, I want to say that for a more efficient, correct roasting, it is necessary to use a liquid flux, which is introduced under the chip.

All the subtleties and nuances in this post are not described, since this is not a guide to action, but an introductory article.

I recommend paying attention to this article about frying video cards .

UPD I recommend reading the article Problems with nVidia chips .