Budget Custom Loop Project

[TTurnerdudEE]

The notion of cheap water cooling has been under scrutiny since well forever, this being that it will break and kill your components and that the performance will be quite frankly rubbish. However, how plausible is building and running a cheap custom loop in the long run? Is it nearly as bad as it's made out to be? In order to find out I've built and fully custom water cooled PC for under $100.00 ( Australian Dollars ) this including 1x cpu block , 1x gpu block , 1x 120 mm rad , 1x pump , 1x res, distilled water and 3/8 id 5/8 od tubing.

 

The system components ( which are modest for obvious reasons ) consist of a Pentium G2020, GTX 560 ti, 4GB ram, 60GB ssd, 320GB HDD, 550 Watt PSU and a modded coolermaster elite 120. While these specs are not overly 'impressive' by any means the point of this project is to determine the long term viability of cheap water cooling rendering the specs of the system mostly irrelevant (except for thermal output). 

 

Before undertaking any installation of water cooling components an air cooled benchmark was a must, this using the stock intel heatsink and the video cards custom cooler ( Gainward dual fan cooler ). This yielded mediocre scores across the board with the cpu idling around 40 degrees Celsius and pushing 60 under load. The GPU similarly idling around 40 but pushing well towards 80 degrees under load. These temperatures accompanied by a small case with limited airflow made for an interesting experience especially during summer, this giving rise to this project.

 

Firstly, the purchasing experience. The general consensus of purchasing anything from China through ebay is that it's going to be a mixed bag, there is no guarantee of quality or the item arriving in one piece at all. My experience with purchasing the items was, contrary to popular belief utterly fine. The items arrived within their suggested arrival date and all in commendable working order. The quality of the items was also note-worthy where everything looked and felt sturdy. 

 

Secondly, the installation. The installation of the components is where the first hiccups arrived, the most notable being the cpu block. The cpu block offers mounting for a variety of AMD and intel sockets, the issue being however as a result of the AMD mounting hardware it didn't fit over certain components on the motherboard ( The board is a micro ITX board so your mileage may vary ). This however was quickly amended with a metal saw and some electrical tape to ensure nothing shorted out on the mounting hardware. Everything else regarding the installation was fine until it came to the radiator, while the radiator itself was fine the ID of the barb fittings to which tubing is to be put over is smaller than it was advertised ( 3/8 ID ). This however was quickly fixed via using spring clips that came with the CPU block where the fit was good enough on the cpu block to not warrant their application. Every barb fitting is secured with a zip tie to maximize the longevity of the loop. 

 

thirdly, this brings us onto filling the loop. Filling the loop was no more difficult than any other system, it filled up quickly ( due to the low volume of water required in the loop ) and like many custom loops only sprang a small leak which was easily dealt with with an additional zip tie. 

 

Thing bringing us onto cooling performance. With running two fans ( push and pull ) on the radiator at a set speed of 1200 rpm the temperatures did see an improvement. The cpu idled around the mid 30's as the did the GPU, the real difference being in the under load scenarios ( 3D mark fire strike ). Under load the CPU runs in the mid 40's and the GPU runs in the mid 50's only ever reaching 60 once. These temperature improvements yield a great performance increase as it allows for significant GPU overclocking headroom where originally overclocking the GPU would not have been possible. Additionally, the noise output of the pump is minimal. The pump is set to run at full ( as it's not pwm ) this being 400l/h and at this rate the pump produces only the slightest hum. 

 

While what I've documented so far does not really cover a bulk of the perceived problems with budget custom water cooling I intend to use the rig daily update this forum on the rig to ultimately determine if such a rig will work reliably in the long run, the only way of validate this is actually doing it. 

 

Concerns people may have:

- Mixing metals : While I'm well aware that mixing metals in a water cooling loop isn't a good idea the point of this project / experiment is to see what sort of performance / life can you get out of the gear and if any corrosion does occur , how long into the life of the loop does this become a problem and warrant a change.

- Something leaking : While in any liquid cooled system there is a possibility of something leaking there is an inherently higher risk in a system such as this using barb fittings mostly supported by zip ties. Again, this system is not intended to be a reliable long term rig, it's an experiment and should be treated as such. 

- Matter growing in the loop: fear not, there is a piece of 99.9% silver to add to the res to reduce the likeliness of something growing in the loop. 

 

I will upload pictures of the system at a later date ( it's currently dark and there is incredibly poor lighting not warranting an exceptional image. )

 

Looking forward to hearing others thought on this concept and weather or not they believe the system will 'stand the test of time'

 

- Tristan

 

 

 

 

 

 

[TTurnerdudEE]

Here is the first selection of images of the system. While it may still look like a work in progress this is because it is. As of currently the systems cooling works as intended there is still a need to properly tent to the cable management of the system. Due to the unique layout chosen over the traditional layout of the Elite 120 this is quite a challenge due to the minimal places cables can be hidden. 

 

As of currently the loop has experienced no issues and is working as intended, idle temperatures are well in the thirties at worst and the loop is remarkable quiet. 

 

Questions people may have:

 

- Why such a strange layout for the case?

Simple, I wasn't overly fond of the 'short and fat' look of the case in it's oem state, I decided the best way to rectify this was to put it up on it's side, mount the power supply where the drive cages had originally been located as to show off the motherboard and allow adequate room for tubing to the cpu. Additionally this layout allows any leakage to drip right out the case away from sensitive hardware and allows me to easily monitor the loop as the lack of side panels ( or any panel for that matter ) gives an unprecedented view into the system ( no cable management options being the only 'downside' in my opinion).  

 

- What is the radiator mounted to?

The radiator is mounted to the power supply. The power supplies fan has been inverted causing the power supplies original exhaust being re-purposed as an intake allowing cooling for the power supplies internals as well as the radiator. 

 

- Is mounting a radiator to a power supply a good idea?

Probably not, I have however taken the best precautions to eliminate coolant from entering the power supply in an 'undesirable situation'. This has been achieved through having the intake and outtake of the radiator at the bottom so if the fittings do leak they don't drink onto or into the power supply.

 

- How did I mount the power supply to the case?

This was achieved through re purposing the original power supply mounting hardware by drilling holes through the shroud and into the original floor of the case, I then put screws through and it is held there sturdily. Additionally I have Velcro holding the opposite side of the power supply up in order to provide additional support.  

 

- How did you plug in the power supply cable?

I ordered a 90 degree adapter from ebay and replaced the original one. I found this adapter to interfere with the front of the case so I cut it down and secure all openings to the cable housing with a thick layer of electrical tape.

 

- How did you make all the cables black ( as poorly done as it is )?

I plasti-dipped all  the cables. I did however do this on a windy and rainy day in 30 minutes thus why the atrocious finish. I intend to cover up all the blemishes with a permanent marker as it's the easiest way to 'fix' the issue without removing the power supply and therefore cooling system from the system. 

 

If anyone has any questions about my atrocious mod or the loop I'll be happy to answer! c:

 

 

[TTurnerdudEE]

Update:

So far into the project there have been no hiccups whatsoever. The loop is utterly fine, quiet, runs like a dream and most importantly has allowed a considerable performance increase. The enhanced cooling for the graphics card has allowed me to overclock the card to 1.01 Ghz on the core, 2.02 Ghz on the memory with a core voltage of 1.08 V. This has provided a significant performance boost in 3D mark skydiver boosting the score from 7558 to 8278 ( 9.5% improvement ). Additionally I've tidied up the system  ( as I assume many of you were horrified by the previous pictures ) but I soon intend to migrate the system to a 'proper' case due to its success and it not requiring nearly as close monitoring as I thought. This will also provide a perfect moment to examine the loop for any corrosion or any un-noticed issues. I will likely be moving the system into a Nanoxia deep silence 4 claw case, this is as I believe it to be a brilliant little case ( it really doesn't need to be massive and having a smaller case will allow me to potentially take it to lans ) and that it's relative affordability. This case choice is not set in stone and any other recommendations would be greatly appreciated c: Once again, if you have any questions regarding the project I'd be happy to answer them! 

[OneshotOtter]

Ghetto, but totally awesome and something I want to do!!!

[KE2012]

Well call me intrigued.  I'll throw this a follow as I'm curious as of the long term outcome but as a guess I reckon your pump will be the component that kicks the bucket first if it dies before you end the experiment.  

[TTurnerdudEE]

On 3/24/2016 at 1:12 PM, KE2012 said:

Well call me intrigued.  I'll throw this a follow as I'm curious as of the long term outcome but as a guess I reckon your pump will be the component that kicks the bucket first if it dies before you end the experiment.  

Yeah I think you're probably right, it's the most 'complex' part of the loop and sometimes oddly manages to suck some air into the loop at startup ( still trying to figure out how it manages this haha ) which is a tad of a worry, other than that it's all going strong c:

[KE2012]

14 hours ago, TTurnerdudEE said:

Yeah I think you're probably right, it's the most 'complex' part of the loop and sometimes oddly manages to suck some air into the loop at startup ( still trying to figure out how it manages this haha ) which is a tad of a worry, other than that it's all going strong c:

Good to hear it's still going, although if it had failed in under 30 days even I would be a litle dissapointed. Anyway how much longer you plan to run the test? And try topping up the res a bit more, I had a simmilar issue a while back with air somehow getting in my loop and this was a solution to mine. 

[TTurnerdudEE]

On 3/30/2016 at 4:04 AM, KE2012 said:

Good to hear it's still going, although if it had failed in under 30 days even I would be a litle dissapointed. Anyway how much longer you plan to run the test? And try topping up the res a bit more, I had a simmilar issue a while back with air somehow getting in my loop and this was a solution to mine. 

Yeah, if it had failed to last 30 days I think the results would have been rather conclusive haha. I intend to run the loop until something dies or something goes horribly wrong. I feel like at that point a benchmark for the reliability of cheap water cooling components can be set. I think you're right about topping up the res, I will give it a top up c: cheers mate!

[Julian5]

On 2/4/2016 at 2:14 AM, TTurnerdudEE said:

Yeah, if it had failed to last 30 days I think the results would have been rather conclusive haha. I intend to run the loop until something dies or something goes horribly wrong. I feel like at that point a benchmark for the reliability of cheap water cooling components can be set. I think you're right about topping up the res, I will give it a top up c: cheers mate!

Hows it going?