I just installed the split water cooling yesterday, and I hope to see yours

[Chinese player Mr. Wang]

＄=Dollar

CPU:intel 10600Kes 5g 1.23v 55＄（Secondhand）
GPU:msi 1070ti 97＄（Secondhand）
mainboard:msi z490 unify 48＄（Secondhand）
memory：asgard Samsung B-die 8G*2 OC4400C16 66＄（brand new）

1、CPU cold head model：bykski 0.08mm，9.49＄（brand new）

2、Graphics card cold head model：MSI GTX 1080 SEA HAWK EK X，6.98＄（Secondhand）
This card has no light slot, so I can't install the light strip, which is a bit frustrating for me

3、Water pumps and tanks：bykski Lift height 5 meters, Flow rate 1000L/H,32＄(Secondhand)

4、Pump in the upper left corner：ZKSJ 24VDC55E ，Head 16 meters，Flow rate 2400L/H,36＄（brand new）.
Not on yet, I added this pump to test the temperature difference at high and low flows. At the same time, I have tested that this pump can reach 14L/min without a heat source and only connected to the copper radiator cycle. However, when I tested the flow rate of 7L/min, the noise was unbearable, which was a big drawback.

5、Semiconductor chilled sheets：TEC1-12703，12V 30W，1.81＄（brand new）。
It is installed behind the CPU cooler in the upper right corner, and it wants to further cool the cooling circuit by measuring the cooling sheet. Unfortunately, it's hard to cool down when the computer is on standby. It may require 10 times the cooling capacity of the semiconductor cooling chip to be effective. But the cost of power consumption and electricity is too high, and I don't want to try again.

6、Copper radiator：bykski，240+360，55.8＄（brand new）.
I would like to say that this copper radiator has a great disadvantage, it has a large amount of flux required for soldering during manufacturing, and it is difficult for our users to clean it after they buy it. When the cleaning is not thorough, as long as 1 to 2 weeks, the inner wall of the transparent water pipe of the waterway will adhere to the white powder, and the bottom of the water tank will also precipitate white powder, this problem has bothered me for a long time, I am very distressed, I don't know if there will be improvement after this cleaning

7、Temperature sharing: When I was playing the PUBG game, the room temperature was 24°C, the water temperature was °C, the CPU was 40°C-45°C, the GPU was 36°C, the flow rate was 2.4L/min, and the pressure was 0.3bar

8、Finally, I wanted to open the lid and remove the cover from the CPU (which I had already done), and use the cold head to make direct contact with the liquid metal, but unfortunately I couldn't find the right fastener, so I even polished the four plastic corners of the CPU base of the motherboard to put the buckle in order to fit it properly. The graphics card also wants to conduct heat and also wants to use liquid metal, but I don't know how to prevent it from leaking. Also, I want to use Coollaboratory Liquid Extreme, but it is not available in China. In this regard, can you give me some suggestions or comments, welcome to leave an interactive message for exchange

Please comment on my water cooling, I hope you can give me some suggestions for further lowering the temperature, such as material, choice, brand. It's too expensive for me to afford, you don't need to recommend it, haha, just the parts of the ordinary civilian consumption level.

 

[ShrimpBrime]

Hello! This is a very impressive build. I want to suggest a better TEC part number that would be more useful when chilling water that is being heated. This will require some additional parts for controlling the TEC because it's closer to 130w vs only 30w each.

12715 data sheet. https://peltiermodules.com/peltier.datasheet/TEC1-12715.pdf

These are extremely power hunger and run extremely hot. The water loop for the TECs must be twice as big than that of the CPU which should be pump tubes and cpu waterblock only and all of it insulated. Will also require two flow rates for each loop, I do suggest to regulate the cpu loop because it's easier to cool water when it's flowing slowly. The TEC Loop at least 3 120.3 radiators for only 3 to 4 TECs. I would also regulate the TECs as they are capable of melting wires.

Since there's a 30c difference on average between hot and cold side, you are maybe 5% efficient chilling water when there's a CPU load of sorts. 10% on a good day. Idle and wait an hour can show off some really nice numbers to show off, but realistically, you are still within a range of where it just doesn't matter, you've gained cpu power efficiency by running cooler, but lost it all trying to chill it this method and are at -15% power usage.

Rate of exchange is slow. Moving any type of a temperature (thermals) is resisted by the material it moves through including thermal pastes between gap of materials.

Knee deep in the TEC chilling probably 10 years ago now, I successfully chilled a 2700X at 0C and less with a single 12715. Water temp using tap to drain to cool the TEC was at 8c. None the less, 4ghz at just under 1.200v. I do not have screen shots handy at this time. This was no easy accomplishment. The TEC water loop actually needed to be chilled. Ice water in a bucket actually brings temps at around -25c idle. OK on top of that, the IHS plate was too small to store any thermals. I tried several plates after a de-lidding of the cpu. Using the largest one I had, produced the best results. I was using the TEC to chill the plate, not the CPU. More copper, more thermal storage. So you can run a load for small periods of time, and remain frozen (to prevent condensation freezing and un-freezing). In that case, you insulate everything anyways. At least with a PGA socket, you can use electrical grease to help prevent pin rot if using for long periods of time (1 year or more frozen and unfrozen cycles) always seems to present problems.

Anyways, I really appreciate seeing your work here! This looks phenomenal! And you have ok results. I like it a lot! Good job!

 

[Raffles]

Good job Mr Wang. Very impressive, I've never seen anything like it

 

[Chinese player Mr. Wang]

Hello! This is a very impressive build. I want to suggest a better TEC part number that would be more useful when chilling water that is being heated. This will require some additional parts for controlling the TEC because it's closer to 130w vs only 30w each.

12715 data sheet. https://peltiermodules.com/peltier.datasheet/TEC1-12715.pdf

These are extremely power hunger and run extremely hot. The water loop for the TECs must be twice as big than that of the CPU which should be pump tubes and cpu waterblock only and all of it insulated. Will also require two flow rates for each loop, I do suggest to regulate the cpu loop because it's easier to cool water when it's flowing slowly. The TEC Loop at least 3 120.3 radiators for only 3 to 4 TECs. I would also regulate the TECs as they are capable of melting wires.

Since there's a 30c difference on average between hot and cold side, you are maybe 5% efficient chilling water when there's a CPU load of sorts. 10% on a good day. Idle and wait an hour can show off some really nice numbers to show off, but realistically, you are still within a range of where it just doesn't matter, you've gained cpu power efficiency by running cooler, but lost it all trying to chill it this method and are at -15% power usage.

Rate of exchange is slow. Moving any type of a temperature (thermals) is resisted by the material it moves through including thermal pastes between gap of materials.

Knee deep in the TEC chilling probably 10 years ago now, I successfully chilled a 2700X at 0C and less with a single 12715. Water temp using tap to drain to cool the TEC was at 8c. None the less, 4ghz at just under 1.200v. I do not have screen shots handy at this time. This was no easy accomplishment. The TEC water loop actually needed to be chilled. Ice water in a bucket actually brings temps at around -25c idle. OK on top of that, the IHS plate was too small to store any thermals. I tried several plates after a de-lidding of the cpu. Using the largest one I had, produced the best results. I was using the TEC to chill the plate, not the CPU. More copper, more thermal storage. So you can run a load for small periods of time, and remain frozen (to prevent condensation freezing and un-freezing). In that case, you insulate everything anyways. At least with a PGA socket, you can use electrical grease to help prevent pin rot if using for long periods of time (1 year or more frozen and unfrozen cycles) always seems to present problems.

Anyways, I really appreciate seeing your work here! This looks phenomenal! And you have ok results. I like it a lot! Good job!

OMG, did you play TEC ten years ago. It can be seen that you are also a DIYPC cooling fanatic lover of good people, haha... The power consumption of the TEC1-12715 you recommend is too high, and I won't try it easily, because it requires additional money and time investment, such as power supply, refrigeration pads, manual disassembly and assembly man-hours, and even electricity costs, which I am more economical to consume. Moreover, the installation of this split water-cooled computer has exhausted me... Maybe after a while, I'll be lucky enough to find a free high-power power supply and I'm bored and want to try it, so I might take it apart and try it. If I try, I'll keep posting and sharing. Thank you for your reply... Having said that, I installed this TEC1-12703 before and then disassembled it after the test run, and found that the middle hollow of the cold surface has no silicone grease, and the hot surface silicone grease is only a little around the middle, and there is no silicone grease around, I don't know if it was not installed correctly before, for example, the two sides of the TEC silicone grease were not completely smeared and laminated, resulting in dry burning part of the TEC during the trial operation, and the TEC was burned out, so the refrigeration was not good.

Good job Mr Wang. Very impressive, I've never seen anything like it

Thank you for your message compliments

 

[ShrimpBrime]

OMG, did you play TEC ten years ago. It can be seen that you are also a DIYPC cooling fanatic lover of good people, haha... The power consumption of the TEC1-12715 you recommend is too high, and I won't try it easily, because it requires additional money and time investment, such as power supply, refrigeration pads, manual disassembly and assembly man-hours, and even electricity costs, which I am more economical to consume. Moreover, the installation of this split water-cooled computer has exhausted me... Maybe after a while, I'll be lucky enough to find a free high-power power supply and I'm bored and want to try it, so I might take it apart and try it. If I try, I'll keep posting and sharing. Thank you for your reply... Having said that, I installed this TEC1-12703 before and then disassembled it after the test run, and found that the middle hollow of the cold surface has no silicone grease, and the hot surface silicone grease is only a little around the middle, and there is no silicone grease around, I don't know if it was not installed correctly before, for example, the two sides of the TEC silicone grease were not completely smeared and laminated, resulting in dry burning part of the TEC during the trial operation, and the TEC was burned out, so the refrigeration was not good.


Thank you for your message compliments

Ive encountered TECs getting earlier failure rates from similar issues where a water block I had used did not fully cover the corners. Wasn't long before the heat made its way to burning some thermal grease. The application must be generously applied.

Yes the 12715 Tec can certainly consume a lot of power, that is why I had suggested making a controller for them. They are recommended 15v application, but I never bothered. Using a standard ATX psu works fine especially for the lower wattage TECs you are using.

I also agree that TEC cooling takes many man hours and additional money over all ambient only liquid loops just for cooling a cpu and or gpu. Ive calculated it takes more than double the wattage to cool a cpu and or gpu with TECs chilling the loop. You must take into account all of the materials that will absorb the thermals. Anything that water touches, will leak your chilling. So the cool down period should be no less than 1 hour before applying an applicable cpu load.

This is one of my favorite cooling methods, strictly because of the challenge. I always have a TEC somewhere in random drawers lol.

 

[Chinese player Mr. Wang]

Ive encountered TECs getting earlier failure rates from similar issues where a water block I had used did not fully cover the corners. Wasn't long before the heat made its way to burning some thermal grease. The application must be generously applied.

Yes the 12715 Tec can certainly consume a lot of power, that is why I had suggested making a controller for them. They are recommended 15v application, but I never bothered. Using a standard ATX psu works fine especially for the lower wattage TECs you are using.

I also agree that TEC cooling takes many man hours and additional money over all ambient only liquid loops just for cooling a cpu and or gpu. Ive calculated it takes more than double the wattage to cool a cpu and or gpu with TECs chilling the loop. You must take into account all of the materials that will absorb the thermals. Anything that water touches, will leak your chilling. So the cool down period should be no less than 1 hour before applying an applicable cpu load.

This is one of my favorite cooling methods, strictly because of the challenge. I always have a TEC somewhere in random drawers lol.

View attachment 405739

Oh my god, you've opened the lid on the CPU too. And you also have the HEATKILLER IV brand cold head, this brand is very rare among our Chinese players, after all, there is no direct distributor in China, and the purchase needs to be ordered across borders, I heard that the price is expensive, so it is rare... 1. I need twice the wattage, which is terrible, when I play games, the CPU is -70W, the GPU is -150W, a total of 220W, don't I need to use 440W-TEC, oh no, it's terrible. 2. My goal is to be very satisfied if I can suppress the water temperature to 5°C at room temperature, such as 30°C at room temperature and 25°C at water temperature. 3. An old player in our Chinese forum said that it is best to turn off all the fans of the radiator, which is conducive to reducing the leakage of cold water and air conditioning in TEC, because the fan blows the radiator into the high ambient temperature. Send you some pictures of the device he's playing, he's playing very, really crazy. I've got to go to bed, it's 1 a.m. Good bye。。。

 

[ShrimpBrime]

Oh my god, you've opened the lid on the CPU too. And you also have the HEATKILLER IV brand cold head, this brand is very rare among our Chinese players, after all, there is no direct distributor in China, and the purchase needs to be ordered across borders, I heard that the price is expensive, so it is rare... 1. I need twice the wattage, which is terrible, when I play games, the CPU is -70W, the GPU is -150W, a total of 220W, don't I need to use 440W-TEC, oh no, it's terrible. 2. My goal is to be very satisfied if I can suppress the water temperature to 5°C at room temperature, such as 30°C at room temperature and 25°C at water temperature. 3. An old player in our Chinese forum said that it is best to turn off all the fans of the radiator, which is conducive to reducing the leakage of cold water and air conditioning in TEC, because the fan blows the radiator into the high ambient temperature. Send you some pictures of the device he's playing, he's playing very, really crazy. I've got to go to bed, it's 1 a.m. Good bye。。。

It's natural for phase change to cost a lot of wattage. This is similar to the idea of a window air conditioning unit. Or perhaps a refrigerator. They are only 2 different methods, but to make cold cost energy, more than half of that energy is dissipated in heat and less than half to make a chilling. And why the heat sink on the hot side of the TEC is very large, just as you need a larger condenser than an evaporator.

Remember you are chilling the radiator, that costs you time and chilling efficiency. The liquid has to chill eveything it touches. So less parts, less wattage required. When I run a chiller loop for the CPU, I eliminate the radiators entirely. Quick connects make installing and removing radiators fairly simple. Something perhaps to consider.

The HeatKiller IV (non pro) is what that block is. I don't really like it, I wish to have the pro full copper block instead. Copper is great for thermal storage. And this helps with chilled loops. If you can keep the temperature gradient lower, that means it takes a long time to chill something, but also a long time to heat it back up after chilling. Water absorbs heat very quickly. TECs that cannot chill the loop will just be heated on both sides. And the reason I mentioned using high wattage TECs. They are much more effective, and great with use with a controller.

Old pictures, not current.

The larger cold plate was 100x more effective than the small cold plate.
Yes this is IHS plate replacement (lid-less PGA_)