Ik heb een site met eeb verslag over phase change cooling gevonden, met een redelijk uitgebreide handleiding en het gebruikte gas, deze gast kwan tot -43 graden celsius !!!
DaBit , wie weet heb je hier wat aan,
de link :
http://cpusite.examedia.nl/sections/steve/its-cooling-time.html
En ik heb de tekst hier in geplakt SUXXES !!!!!!!!!!!
The Project
I had read just about everything you could imagine on cooling systems for computers, looked at all the "projects" posted on the web sites, and looked at both KryoTech and Asetek and found nothing for dual processor systems that would go really low temp. Were talking sub 50C temps.
The Kryo Tech and Asetek both claim "potential" 40C but because of our environmentally friendly R-134a freon, this is pushing the outer limits of the efficiency of this freon. These systems might achieve 40C under ideal, no load conditions, but could never maintain it under an actual maxd out processor load, and it is definitely not the temperature within the core of the processor.
I had three obstacles to overcome, space constraints, two processors and commercially available components. The system had to be compact, it had to cool two PIII processors and everything had to be available over the counter or on the internet.
Because of the desire to achieve sub 50C temps, I determined that this was a two stage approach. I could achieve exchanger temperature of 30C to 35C with a freon based cooling system but would have to resort to Thermal Electrics (peltiers) to bring it down to the target temps.
Part 1: The 1st Stage Liquid Cooler
After reading much material on the "phase change", freon based cooling systems, I had learned that the efficiency of the system was controlled heavily by the size and thermal properties of the condenser. I went to local A/C supply stores, A/C shops and looked everywhere on the Internet for a small, compact condenser coil. I found this unit at a recycle shop for a little refrigeration unit. ( If you look at the Kryo Tech and Asetek units, this look like the same unit).
Now that I had a clear understanding of what would make my system really cook, it was apparent that this was a do-it-yourself package. I went to a local A/C shop and bought a used condenser coil that had a punctured tube for twenty bucks. Out comes the Dremel with a cut off wheel and wallah, the size needed.
This little puppy gave me three rows of 6 tubes each. Assembly was straight forward, a little solder, some copper tubing, a couple of pieces of sheet metal and two high CFM fans (bought at electronics clearing house for $2 ea.)
One of the hurdles encountered with the use of these little refrigeration units is that the way they control temperature is by cycling the unit on and off. For my application this would cause an unacceptable level of fluctuation in temperature at the exchanger. Once I achieved the desired cooling temp, the unit would cycle off and the processor would add heat faster than the unit could cycle back on and recover the temp gain. Plus with the processors sitting idle, the heat generated would be substantially less than at full load. What to do? As you can see from the two pictures, there is a brass gadget attached. This is an adjustable hot gas bypass valve. It allows me to control the "efficiency" of the condenser, and keep the final temp constant without cycling the compressor. It also doesnt let the temp continue to drop under no load until the evaporator "freezes-up".
Now for the compressor, after a shit load of research, I found the Tecumseh AZ0349Y compressor. This little baby ($122.00 Retail-aahh!) is designed as a constant duty, non-cyclic, R-134a unit, capable of displacing 1655BTU/hr with the right condenser - evaporator combination.
Now after all the material I had read about the efficiency of the R-134a freon, I thought about converting the unit to R-12. That was a mistake. The oil is not compatible, the seals are not compatible and you need an act of congress to get the R12. Then I found
http://www.epatest.com/qwikboost-press-release.html Quik-Boost® additive. This increased the efficiency and cooling capacity by 10% to 15%, making it a little more equal to R-12.
The next decision was the evaporator. Do I couple direct to the compressor and allow the "phase change" to occur at the processor or cool a liquid and circulate it though the exchangers? Again, after all the research and reading, I determined that I would build a small chamber to cool a 40/60 mixture of glycol and water. I decided this because the compact size of the exchanger on the processor would again lead to fluctuation in temp if I had the refrigerant "phase change" at the processor. Surface area, volume, material all effected the decision. The use of a liquid, along with the fact there would be a reserve in the tank, would result in more stable temperature at the exchanger and would allow immediate boot of the machine because you didnt have to wait for the compressor to "wind-up" and start dropping the temp. The water naturally had a "reserve" of heat storage capacity until the compressor became efficient.
With the fact that size was a restriction, I went on to fabricate the evaporator and coolant chamber. Using 20 gauge sheet metal and solder, I built the container.
After soldering the sheet metal together, I lined the inside with about 50mils of liquid urethane for that "extra" protection. The evaporator coil was constructed out of Œ" copper tubing and I added a large copper "bulb" to increase the efficiency of the evaporator. In my hand is the Aqua-Jet 404 aquarium pump. ($12.45 at the local per store). Capable of 108GPH, and as you can see, very small, it was placed in that little area above the evaporator coil. This little chamber held the evaporator, pump and 1.2 quarts of coolant. Since we would be looking a 25C on the coolant, it had to be insulated.
I built another 20 gauge enclosure and filled the area between them with high density foam insulation. Everything was mounted and refrigerant piping completed.
Initial testing began. Not knowing what the capacity of freon was, it was trial and error for adding freon. After a half a dozen fills, the most efficient temperature was achieved at 9 oz. of R-134a and 1 oz. of Quik-Boost®. I measured the amperage draw of the compressor and compared it to the temp at the evaporator. I also measured the temperature of the compressor itself. Finally the set temp at the evaporator was 24C. This required 145W of power and the compressor never went above 54C.
Stage 2: The Exchanger and TECs
When you consider the use of TECs (peltier elements) you need a power source. The more heat you want to move or the greater the temperature change you want requires more power. Since I was dealing with two processors and wanted up to a 40C additional temperature drop, I knew that a typical computer power supply wouldnt cut it. Out come the power supply design handbook and after many trips to Radio Shack, Sandys Electronics and the local swap meet, I had a 35amp, 15volt, variable power supply.
Making it variable would allow me to adjust the capacity of the TECs to the point they would displace the greatest temperature with the least amount of parasitic heat pumped into the cooling system.
Building the heat exchangers would become a challenge. They had to be rigid in order to ensure a truly flat surface to mount the TECs. I ordered a piece of Œ" x 2" x 4 copper bar stock ($32.54). This would become the exchangers and the thermal plates mounted between the processor and TECs.
After using the Aztec® program downloaded from Melcor @
http://www.melcor.com , I decided on the CP1.4-127-LH1 peltiers with two mounted in series on each processor. This would give me the ablilty to move 34W of heat with a delta of 40C. Since the peltiers were to be wired in series, it would allow me a extreme range of power control since they would begin to move heat at 2V and would no longer be efficient at 17.7V. These were sandwiched between the exchanger and the thermal plate using Cool-it-Dude thermal grease.
Now that I had the "parts" together, load testing began. I decided to use one of the peltiers to "induce" a 50W load on the thermal plate and just wired up the two others to a standard computer power supply (no control). This resulted in 34.6C.
I was now ready to build the insulated enclosures for the processor/exchanger assembly. The testing was exciting in the fact that with a 50 watt load (twice that of the PIII @ 100%), I had achieved 34.6C. With just a little math, -60C was realistic if I could insulate properly and control the condensation. Off to the local hobby shop. I purchased .060, .080 and .120 inch styrene sheet stock to build the enclosures. 3/8" high density foam insulation was also bought from my local A/C shop.
Stage 3: Putting it together
Now that all the parts are together and the system testing complete, its time to move to the computer and install everything. I am using a California PC Super Server case and have installed a SmartUPS 1400 internally with 4 33ah batteries. The 1st Stage Cooling system is mounted above the battery rack.
Plumbing was simple but due to the space constraints, insulating the line was a little more involved. Since the coolant is 25C, condensation is a big concern so insulating everything was a priority.
The true test, a working computer. Everything was turned on and the power supply to the peltiers was set at 12v. Before I had the login screen for WinNT4.0 the processors had dropped to 4C. Within twenty minutes I was at 28C. This was getting exciting. I loaded 3D f/x and started rendering a 2min animation (1.23Gig file) which takes about 2hrs 3min. The system continued to drop until I had reached 43.1C (measured from the center of the thermal plate). Once I finished the rendering, after about 15 minutes the processors were at 58.3C and 59.7C. I set the power supply to 9.7V and maintain a temperature of 35C right now and will continue to test different voltages and check for condensation to be certain that everything is air tight. Next stage will be to cascade the peltiers to see if 80C can be achieved. This will get me real close to the ability to double the clock speed. (When the K7 comes out, this will no longer be an Intel machine. I am done with those prima donnas.) By the way, the extreme low temps did improve my WinStone and WinBench Scores by about 12% even through the Windows Diagnostics still shows the FSB @ 100Mhz even thought it is set at 112Mhz in Bios. This lets me to believe that the locks are effected by temperature (some sort of resistor setup)
:strip_icc():strip_exif()/u/32357/groen.jpg?f=community)
:strip_icc():strip_exif()/u/6185/redoct.jpg?f=community)
.
), maar ik denk dat je dat gas wel kan bestellen in alle ons omringende landen (vooral Engeland heeft veel airsoftwinkels). Ik neem aan dat dit hetzelfde spul is als wat je zoekt voor je phasechange- of heatpipesysteem?:strip_icc():strip_exif()/u/43/crop584688f8794de_cropped.jpeg?f=community)
:strip_icc():strip_exif()/u/36877/sq_175.jpg?f=community)