Large Reservoir/Heat Sink Project

[PaulZeN]

I haven’t been able to find any helpful posts on large volume reservoirs, particularly for my recent build, so I thought I would post this to help someone in my position in the future. My first was 15 years with a 20lt plastic container with a pump in the container. My current build is a pump/waterblock combination into a radiator in a 30lt plastic bin, sitting in a 55lt metal bin.

What?
A large volume reservoir is a system that transfers the heat into a liquid filled container either by recycling all the liquid or with a closed loop that runs into it. The reservoir is a heat sink eliminating the need for fans. The container with the reservoir will also radiate heat, so never gets hot enough to overheat the CPU. It is designed for computers that aren’t run 24/7, with the volume of the reservoir being at least 20lt.

Why?
No fans, nice and quiet with no quick temperature increases of the water due to load. It is also simple and cheap. Best cooling possible for certain applications.

Why not?
A large container of water in your house, which could leak, and it takes up space. Not for 24/7 use, but in a cold climate with a big enough container it should work. Looks horrible!!

Parts for current build:
Apogee Drive II integrated pump & waterblock
2 fan radiator (could have used single fan)
3 metres of hose
7 fittings
Irrigation elbows (to create low point, so coolant drains into bucket and not computer if there is a leak at CPU)
30 litre plastic bin (to stop corrosion between radiator and bin)
55 litre metal bin (outside radiator)
Tube of silicon (to seal seams of bin – yes, it did leak)
4 litre bucket (for coolant leak collection)
1lt of coolant

Problems:
RAM radiator fin had to be trimmed to fit water block.
750w power supply would shut down when pump connected, so used another computer power supply - currently powering car sub as well.
Metal bin leaked, sealed with silicon.
Radiator was placed in metal bin and stripped galvanising off a section of bin overnight, so radiator moved into plastic bin. (Yes, I feel stupid now; a copper radiator in a metal bin…)

Performance:
Excellent. The only purpose is for a DAW program to reduce the ASIO driver load from 100% with audio glitches. The cooling improved ASIO 100% to 95%, with overclocking down to 65%.

4770K (1.2 v, 100 x 44.0) runs at 35 to 40 C with minimal use, and during 100% load testing around 70 C. Current weather min 20 C, max 31 C.

After 9 hours of continual use with CPU at around 30%, the water in the plastic container is bath warm, with the metal bin water much cooler. The water cools to room temperature in less than 2 hours.

ASUS Fan Master reduced the pump speed to minimal, so it is never heard at any CPU load.

Next time:

I will use poly irrigation pipe similar to solar heating for pools in a metal container.

 

[Silver_Pharaoh]

My only real concern is bacterial growth. Personally I would not add any sort of PNH Nuke or the like (PT Nuke) as that will evaporate into the air you breath. Same goes with any sort of dye. You DO NOT want to breath that stuff in!!

Other than that, 10/10 for creativity. I love it!!

 

[Mandrake4565]

My only real concern is bacterial growth.

Other than that, 10/10 for creativity. I love it!!

Agreed, good stuff.

 

[sporktar]

Very cool, i love unique "outside of the box" builds like this. Will the radiator really cause corrosion with the metal bin? It seems like the plastic bin you have inside of there is going to insulate a lot of the pc radiator heat from reaching the metal bin, I'd guess it's hurting performance significantly.

 

[PaulZeN]

The bin has a lid on it and only has tap water, as does the plastic container, so it is like a water tank. No food or light for bacteria, but time will tell. Only coolant in the closed loop so no fumes.

The radiator and bin corroded over the first night with an area of the bin stripped of galvanising the shape of the radiator it was close to. Lucky the bin leaked and I discovered it. I was surprised.

The radiator isn’t happy sitting in the water at the moment with some sort of sediment forming. My guess is that the aluminium and copper in the radiator itself is reacting with the water. I will let it fail, and start working on a poly pipe set up to go directly into the metal bin which will improve performance dramatically. I may be able to go down to a smaller bin.

The Y shaped connector failed today with coolant escaping into the 4 lt catchment bucket when the closed loop sucked in a big gulp of air when turning off computer and pump going to full speed before killing all power to the system. The low catchment point worked as designed. Replaced the Y connector with 13 mm irrigation T and 15 mm clamps.

 

[Mandrake4565]

You may be better off with distilled water and a bit of automotive antifreeze in the garbage pail. A suggestion of a cheap bin that you can seal up, is a kitty litter bucket. I use one for my H2O benching setup. Though, I use the cold winter air to help get the water cool.

 

[knoober]

Whats to stop you from filling up an empty 2-liter soda bottle with water and freezing it, to use as a re-freezable ice cube? You could just keep a stock of those in the freezer and chill the water with that right?

 

[Mandrake4565]

Whats to stop you from filling up an empty 2-liter soda bottle with water and freezing it, to use as a re-freezable ice cube? You could just keep a stock of those in the freezer and chill the water with that right?

Yup, just need to take condensation into account.

 

[PaulZeN]

My first was a 20 lt water container with some small holes for the hose and cables for the pump. It worked well and had distilled water and automotive anti corrosive added.



When I took it apart a year later the water block was very corroded on the inside and kind of freaked me out as at some stage it would have failed and fried my computer. This is why I have the closed loop with the commercial water cooling coolant.

The 20lt container warmed up much more than my current set up, being plastic and smaller in volume but still worked well. Not keen on buying 20lt of coolant and dealing with fumes.

No winter to speak of here unfortunately. I thought about ice blocks in containers but not much gain and too much messing around.
With gentle overclocking to reach a required performance, warm water is enough. Going to extremes could be different.

 

[PaulZeN]

Update Number 1: 16 Days Post Construction.

Few things to remember before reading this:
I’m happy with current performance, as I use around 30% CPU load for my DAW
I accept that the following may cause doubts and be disbelieved.

The radiator situation was starting to play on my mind so I had a plan to buy some 4mm poly pipe and calculated that 20x 50cm lengths would still allow me to use my 1 lt of coolant in the system and provide maximum heat transfer straight into the metal bin.

However I convinced myself that 1 m of 13mm should transfer enough heat. I bought what I needed and tested the heat transfer ability of the 13mm when I got home and had serious doubts. Filling a small length with hot water and placing it in cold water reinforced the claims of how good an insulator plastic is. But what kept me going is knowing that it will be sitting in cold water.

Despite accepting the ordeal of draining the coolant, attaching another ‘radiator’ and removing the air bubbles, I went ahead and tested it. Amazingly it worked, so I drained the water, removed the plastic bin and installed the PVC pipe radiator array. The original radiator was seriously corroded and wouldn’t have lasted much longer.

My first test with AIDA64 had the temps go higher and the computer shut down by RealTemp with a peak of 90 C which is where my alarm is set. I restarted and let ASUS Fan Xpert retune and ran it again. I only have the demo version so this is the result after 16 minutes. RealTemp claims the CPU speed of 104.01 x 44, but ASUS AI at 100.0 X 44. Anyway one of AIDA64 test showed higher temps at 104.1 than 100 according to RealTemp when it changed mid test.

First Run

Second Run

Prime 95 and OCCTPT caused the BSD, which my guess indicates that 1.21 v is a bit low. But I’m happy at the moment with the system when I need it.

I’m seriously surprised that 90 cm of 13mm poly is enough, I made a mistake and cut two 5 cm pieces I couldn’t use, to take the heat away, even though my temps are about 5 C higher on idle, but still able to maintain good temps under high load with a bit of extra from the pump in the waterblock, which is now occasionally heard.

I will run this for a while and add some more 13 mm poly, mainly in place of the tubing that connects the poly array, at some time when I can face the ordeal of it…or when I need more power.

 

[sporktar]

Have you considered buying some copper tubing and coiling it in the bottom of your res instead? It shouldn't corrode, as you already have copper in your loop, and the heat transfer would be exponentially better than using plastic style "radiator" bits.

 

[PaulZeN]

Copper tubing would strip galvanizing off the bin.

If I could find an affordable copper bin to use as the heat sink it would be perfect; copper everywhere.

I need a metallurgist to work out the right combination of metals and systems to prevent corrosion.

The plastic works better than I thought, and will stay for now with a few more pieces added in time. For cost effectiveness it is amazing.

 

[PaulZeN]

Update Number 2: 32 Days Post Construction.

After days of musing over the copper pipe, I installed 3m in the metal bin using a plastic garbage bag to stop the corrosion between the copper and the galvanising.

It has now been 2 weeks with no problems and the performance is outstanding. I could have used more copper pipe as I haven’t used my 1lt of coolant in the closed loop yet, but my guess is that I’m at close to 100% return to reservoir water temperature to the waterblock regardless of load.

I installed a temperature gauge in the water and was surprised that it is closer to the maximum range of daily fluctuations even first thing in the morning. The water temp goes up due to the air temp during the day as well as the heat from the computer, but not by much. I still haven’t done a full day of high use, but guess it would be hard to push more than 5 C above ambient.

After 2 weeks there is a thin film of water between the plastic bag and the garbage bin, but it is still firmly forced against it. My guess is that the metal bin will need replacing every year at the regular system service time.

I experimented with 3lt of ice in a plastic container. I left it overnight and it wasn’t thick with ice, but still dropped it in. The water cooled and slowly climbed giving me around 3 hours before the initial temperature was reached. Not sure of practical use, but may allow for higher OCs that are temperature sensitive to be used on warmer days. Next time I will wait until it is frozen solid and see what it does on a high use day. It may need some insulation so it cools the reservoir water slower.