ENCYCLOPEDIA OF RADIO ELECTRONICS AND ELECTRICAL ENGINEERING Water cooling system for computer. Encyclopedia of radio electronics and electrical engineering Encyclopedia of radio electronics and electrical engineering / Computers Recently, computer components are getting hotter, in order to cool them, more efficient cooling systems have to be used. Despite the reliability, simplicity, and low cost of traditional air cooling systems, air has a low thermal conductivity, which means it is not the best solution for cooling, not to mention the noise that is created with a large air flow. The latest coolers emit just a fair roar, while the efficiency becomes acceptable, but not always sufficient. Recently, water cooling systems have become increasingly popular. For those who look at such an idea with irony, I will say that it is not for nothing that this method is used in the automotive industry, remember the notorious air-cooled Cossack, which, despite a weak engine, could not cope with cooling. Water removes heat much better than air, which is why water systems are much more efficient than air systems. It was the noise of the computer that inspired me to create water cooling. Firstly, I work at night, and the computer is next to the bed, so it prevents my wife from sleeping. Secondly, a long stay in noise (and I sit at a computer> 12 hours a day) does not affect well-being in the best way. Thirdly, it is difficult to overclock the system (and I need a productive system for video processing) using standard cooling means. Today, there is a lot of material on the Internet on ready-made systems from leading manufacturers, as well as on home-made cooling systems (home-made systems are usually more efficient). I don’t want to repeat myself and make another review of water cooling systems (I’ll give them in the links used), based on my experience, I’ll just try to describe my design beautifully and in detail. We will consider the main components of the water cooling system on the model of my design: Water is supplied through the hose with a pump (3) to the water blocks (5,6,7), passing through the channels of the water blocks, the water takes heat, and then it is fed into the radiator (2), blown by the fan unit (1). Thus, heat is forcibly removed outside the case. The expansion tank (4) prevents the presence of air in the system, increases the volume of coolant (water) in the system, thereby increasing the inertia of the system, besides, the water in the tank cools the pump. The inertia of the system in our case is just a big plus, because the processor, as a rule, does not work constantly at full load. While it is running, the system heats up smoothly, and during the idle time of the processor, the water has time to cool. I note that this rule does not always work, for example, on my 3200th processor, 1.5 hours of video is encoded for 7 hours, while the processor is loaded at 100%. Fan unit Designed to cool the radiator. Looking at a block of four fans, people immediately have doubts about the silence of the system. The fact is that all fans are turned on through resistors, or you can turn them on not at 12 V, but at 5 V (some may start badly), or at 7 V. At minimum speed, you can’t hear them at all. In the second version, I put a 3-position switch, the first position is +12 V (full turns), the second through a 20Ω 2W resistor, the third two 25Ω 2W in series (4W). This made it possible to turn on the fans at high speeds during prolonged computer loads in the summer in the heat. I’ll make a reservation right away, each mode reduced / increased the temperature of the processor by 1 degree, water by 2 degrees. The difference is negligible, but the noise in the maximum mode is sufficient, but when it is very hot, you can go for it. Guess what the fan tray case is made of. From plastic rulers! They were cheap and were famously glued with dichloroethane. Since it was necessary to output 12V to the fan unit, a relay for turning on the pump was also built in there (pictured). Radiator Cools water by transferring heat to air. My opinion is that the place for the radiator is at the top of the system unit. Inside, it cannot be cooled by already warm air (and the efficiency of such a system VERY strongly depends on the ambient temperature), it is poorly ventilated from below, an independent air flow is not created from the side, and it is very cumbersome in a separate box. According to the material of the radiator, copper conducts heat well, but gives it off poorly, therefore, in this regard, aluminum should be better, but I did not check it in practice, I immediately bought aluminum, besides it was cheaper. The stove radiator from the VAZ-2101 miraculously fit the width of the InWin case, in addition, the threads for the fittings were very quickly and coolly cut in the plastic fasteners of the outlet / inlet pipes. I did not put it in a casing, as ventilation is better, the only thing I pasted strips of porous rubber on the bottom, this raised the radiator above the case and created a better air flow. It does not spoil the aesthetic appearance, but it is very convenient to vacuum the dust :). It would be necessary to put a filter, they say ordinary women's tights give a good effect - cheap and cheerful, as they say. water pump The heart of the entire cooling system is the water pump, which pumps water. The reliability of the water cooling system just depends on it. Medium-capacity submersible aquarium pumps are most commonly used, they are cheap and most common. Their disadvantages are: power supply from 220V, large vibration and limitation of water temperature to 35 oC. Rumors about the unreliability of aquarium pumps are fiction, there are simply poor quality pumps from negligent manufacturers. Choosing a good pump is a problem, branded ones are expensive and it turned out to be not so easy to get them. In our country, for example, the whole city is littered with cheap Polish AquaELs, the reviews about them are not very foresty and their characteristics are strange. The German IHEIMs are considered the best pumps among the people, but they are very expensive. The main characteristics of the pump are performance (liters per hour), and the height of the water rise (meters). The performance strongly depends on the level to which the water rises (the performance without taking into account the rise is indicated in the characteristics), for example, 700 l / h at the zero level turns into 300 l / h at the level of 30 cm, further it is even worse. For normal cooling, a capacity of 150l / h in the assembled system is quite enough (all components in the system reduce the performance of the pump). For the experiment, I bought an inexpensive Chinese RESUN, model SP-1200, capacity 700 l / h, water lifting height 0,8 m, power 12W, dimensions 130x52x109. After checking the performance of the pump in the bathroom, I was convinced of the veracity of the words of the Chinese :) You should not chase powerful pumps, the larger the pump, the more it will heat the water in the tank. For the Inwin A500 case, the height of the water level was 35cm. Expansion tank I bought just such a container for the refrigerator, with a volume of 2.5 liters. She just chic became in the bottom of the case, not even interfering with the hard drives. Its dimensions are 175x175x125. First you need to modify the lid - build in the filler hole and the hole for the cord. I made the filler hole from the top of a plastic vitamin bottle, and for the cord I made a rubber stopper from a medicine bottle. It would also be nice to have a drain hole. Next, we make holes for the fittings in the container body, it is better to do this by heating the tool on fire. The fittings are fastened with nuts on the inside and coated with sealant on both sides. A little about sealant. Plumbing transparent silicone sealant was not the best option, it is better to use automotive sealant, white, opaque, it has much better adhesive properties, more elastic and stronger. The pump must be protected from the walls with a thick layer of foam rubber or suspended, otherwise the vibration from the pump will make your sensations indescribable. After the first experiment of connecting the pump and fitting with a thick silicone tube, I was confused. The rumble from the operation of the pump was terrible, the entire container vibrated, vibration could only be transmitted through a connection with a fitting that was rigidly attached to the body. It was necessary to look for a solution for a soft pump-fitting coupling. The second experiment was carried out with an enema as an interface, because of its volume it had time to dampen the vibration of the water flow ejected by the pump, which created part of the hum. The result was much better, but I could not really fix this design, besides, it hardly fit into the container. The third experiment was more successful, in terms of the combination of construction and silence. I took a white foam material, the one that is placed under the motherboards in the box, and made a tube out of it in 1,5 layers, gluing Moment with glue, and smearing the seams with sealant. But the thickness of the material still gave a slightly harsh design. Then I took the same material, only thinner (various devices are packed in them). By gluing it in 2 layers, I got a very soft and flexible tube. The noise during operation disappeared, there was a faintly noticeable rumble. It was a victory! :) But not long :( After 30 minutes of work, the tube fell off, because the material stretched a little in the water. I had to use clamps on both sides of the tube, and since they could rust in the water, I completely filled them with sealant. The design turned out to be reliable, soon a year and everything works like clockwork. Liquid in the tank. As a rule, distilled water is poured in, if you fill in ordinary water, it will quickly bloom, all components in the system will stop working efficiently due to the plaque that has covered them, and the pump will fail altogether. For reliability, you can add vodka, alcohol, or car coolant to distilled water (the best option). The fact is that antifreeze is neutral to aluminum and copper, with such a liquid you can combine these two metals in the system, and with ordinary water they form a galvanic couple. People say that the best proportion is 1:3, I got 1:4 - 2 liters of water and 0.5 liters of imported antifreeze. For 7 months of operation, the system remains clean, despite the fact that I have used both copper and aluminum in the processor water block. And at the beginning of the experiments, ordinary water was poured into the system, in a week everything in the system was covered with a slimy coating. In the case of adding antifreeze, you must be careful with leaks. After a rebuild, my processor water block leaked after 5 hours of 100% load. After washing and drying, the video card refused to start, even washing it in vodka did not help, I had to buy a new one. And a week later, she miraculously earned. Oh, I still wanted to buy a new one :) Waterblocks Water blocks are the working tool of the system. Perhaps the most difficult part in the system to manufacture. As a rule, it is made of the most heat-conducting material in order to most quickly transfer heat from the chip to the coolant (water). The cheapest of the best thermal conductivity materials is copper. Slightly better than it - silver, two times worse - aluminum. Finding copper ingots of this size is a difficult task for many. I also suffered for a long time in search of material, by the way, it took the longest time. But then I found a place where you can find all the components for the system in abundance - this is, as we call it, a "field of miracles" or flea markets where people sell what they have been lying around from old times. There are an awful lot of options for water block designs. On the Internet, conference threads about the most effective design go off scale for a thousand pages. In general, the design of the water block does not greatly affect the temperature of the processor, but sometimes a couple of degrees is important. I was interested in trying several options in different parts of the system. I don't pretend to be the best waterblock", I just chose the designs I liked the most. Here are these beauties in the project :) Processor waterblock The most important waterblock in the system (center in the photo). The processor, as the hottest device, requires better cooling. I chose a "spiral" type design for implementation, cold water enters the central part of the block, when it hits the base, turbulent flows arise, which increase the heat removal from the metal. The design requires factory production, they made it for me on a machine with a CPU, but some craftsmen manage to do such things on their knees with a drill. I will say right away that I do not like designs on "snot", so I am not a supporter of such manufacture. I will give another processor water block, which occupies the first place in imported reviews, besides, I like it both in design and purely aesthetically.
Under the central fitting there is a so-called accelerator (photo on the right), which enhances the flow of water precisely to the central part of the block. The kit includes 5 accelerators with different slot widths, you can choose the best one for yourself :) Cool thing, everything is done correctly, besides, here's a flow splitter for you, it's a pity that it costs like the whole cooling system :( Chipset Waterblock The chipset is the "coldest" chip in the system of all. With a passive radiator, it will heat up to 40-45 degrees, it would be possible not to put a water block on it at all, but if the heat is removed outside the case, then it should be removed, and this should increase the reliability of the system. I got the simplest water block for it, it is completely (except for fittings) made simply and quickly. Two 10mm thick Plexiglas squares glued together form a lid. In one part, with the help of a drill and a file, a through track of the "snake" type is made, in the other it is a cover with a mount for fittings. The cover is fastened from below along the entire perimeter with M3 screws with countersunk heads. I wanted to make such a block for video, but I thought - plexiglass with GPU temperatures ... As always, it is better to fill all the cracks with sealant. Fitting for water blocks They will have to be sharpened, which also limits the speed of manufacture and increases the cost of the structure. The best material is brass, it is less susceptible to oxidation and corrosion, in addition, it will not conflict with the copper base of the water block. My first fittings were made of aluminum, they are good because they are very light and it is easier to get material for manufacturing, otherwise their advantages end. Tubes Silicone tubes with a diameter of 10-12mm are sold in abundance in car markets. Less - the hydraulic resistance increases significantly, the pump is heavily loaded, its performance drops. More, as a rule, does not allow free space, which should still remain after pushing the system inside. Some are reinforced and some are not. Reinforced ones are good because they do not break on bends, they are bad because they are about 2 mm thicker. It is very desirable to clamp the tubes on the fittings with clamps while the water is cold - the tubes sit tightly, but when the water heats up, water can also leak, so it's better to play it safe. The connection of water blocks can be serial, parallel and parallel-serial. Experience shows that parallel connection does not bring any tangible benefits, but such a system has several disadvantages. The first is the need for additional parts - splitters. Secondly, branched circuits can have different hydraulic resistance and different levels, in this case, in a circuit with less resistance, water will flow with a larger flow, and in another with less. Do we need it? Copper tubing for construction can be found at construction markets that sell plumbing fixtures. Unfortunately, all plumbing "prichindala" have a diameter of 14mm, for me it's a lot. But, after a long search, I was able to find a 10mm tube. With corners under 10mm, everything turned out to be more complicated, I still haven’t found them, but they were, they just stopped being imported (in our country, at least). Tests Test configuration: Board : EPOX 8RDA3+ rev 2.0 - nForce2 Processor: AMD Burton 2500 Air = 24OC CPU standard mode 1820MHz=2500+, Vcore=1,65V Simple + S2Clk:
CPU Burn load:
Load SandraBurn-in Wizard" CPU Arithmetic & CPU Multimedia Test:
РCPU overclocking mode 2200MHz=3200+, Vcore=1,8 В Simple + S2Clk:
CPU Burn load:
Load SandraBurn-in Wizard" CPU Arithmetic & CPU Multimedia Test:
The maximum temperature reached when working in real applications (corresponds to the load results in Sandra) - 45OC. Minimum when the fans were running at maximum speed, at the same load 40OC. The noise at maximum fan speed is approximately the same as the noise of one Titan CU5TB, with it the processor temperature reached 55OC. Total minus 15OC. If we take work with minimal noise, then the difference is already 20OC, but the noise from the CU5TB was still disproportionately higher. Publication: cxem.net See other articles Section Computers. Read and write useful comments on this article. Latest news of science and technology, new electronics: Artificial leather for touch emulation
15.04.2024 Petgugu Global cat litter
15.04.2024 The attractiveness of caring men
14.04.2024
Other interesting news: ▪ World's largest electric skyscraper to be built ▪ Submarine powered by lithium-ion batteries ▪ Tasmanian Devil's Milk Contains Powerful Antibiotics News feed of science and technology, new electronics
Interesting materials of the Free Technical Library: ▪ section of the site Data transfer. Article selection ▪ article The guard is tired. Popular expression ▪ article How did puppet shows begin? Detailed answer ▪ article Cutting teeth on canvases. home workshop ▪ article Piezo sensor in burglar alarms. Encyclopedia of radio electronics and electrical engineering
Leave your comment on this article: All languages of this page Home page | Library | Articles | Website map | Site Reviews www.diagram.com.ua |