AD's water cooling experiment

[AvalancheDiode]

Finally, all parts have arrived today.




So, this is the whole kit:

Motherboard: HP Asus P5LP-LE motherboard from an HP desktop

CPU: Intel Pentium D 830 3.0GHz

memory: some DDR2-667

PSU: some crappy 400W

Monitor: HP vs19

GPU: Intel IGP on 945G.

Water pump: Micro-Jet fish tank pump. Max flow rate: 117gph/440L/hr. $17 bought from a pet supply

Tubing: 1/2ID 5' and 3/8ID 5' $5 total from a pet supply

Tubing connectors: 3/8 to 1/2 barbed (not in the pic) 3/8 Y-connector, barbed (not in the pic as well) $12 for more than I need from McMaster-Carr.

Water block: Zalman WB4 Plus $29.99 from newegg

[AvalancheDiode]

Re: AD's water cooling experiment

Water loop hooked up. Now doing leak test.




OK, I have to explain why there is something missing in the system. I didn't buy that thing because I didn't find cheap ones. I did find cheap ones couple days ago by typing "Transmission fluid cooler" on eBay and Google shopping. I will buy one and mod for computer use if I am serious enough to build a water cooling system for long. However, this is a short test, so I'd stick with large water body for now.

The motherboard was pulled from my HP computer (now rebuilt with current setup on the left). It has no temp monitoring for CPU, and there is no overclocking option. So a thermometer that is used to probe roast turkey is used to monitor reservoir temperature.

[AvalancheDiode]

Re: AD's water cooling experiment

IT'S ALIVE!!!




Room temperature is 84F during the test, and relative humidity is 55%.

Run with pump's full capacity first. For some reason, Prime95 doesn't work, so I ran HyperPi mod instead. It completes 8M HyperPi after 8 minutes. The water by then soured to almost 100 degree. The fan speed remained low. Actually, I've never heard that fan run so low as it is today with that Pentium D.

BTW, don't be fooled by OD of these tubes. The inner diameter is 3/8 and 1/2. These are fishtank tubes which temperature insulation is important. Coolant tubes used in computer cooling are usually more expensive Tygon tubes which are thinner. Tube clamps are not needed partly because of the thickness of these tubes.

This is the pump running in full capacity.




I, for one, is surprised that this pump is made in Italy. That has made my day, though Italy is not none for cheap quality products. Anyway, it has flow control that allows user to reduce flux to 1/3 of maximum flow rate. So I did it to see how bad it would be at very low flow.

This is how low the flux is. There is actually condensation in the 1/2" tube!!!




It still boots and runs fine. HyperPi 2 Process 8M passed without a problem. The water temperature didn't rise as much as when it was at full flux. However, temperature of water block is higher.

[.Gunfire]

Re: AD's water cooling experiment

So is that a temporary reservoir?

[=98=]

Re: AD's water cooling experiment

How about a full list of temps. But, nice layout

[AvalancheDiode]

Re: AD's water cooling experiment

So is that a temporary reservoir?

The whole setup is temporary. The one-gal milk bottle is too small for this purpose. You need 10 times more water and some metal tubing if you want to adopt none-radiator water cooling.

How about a full list of temps. But, nice layout

I'd like to, but this board does not support temperature monitor software. None of the thing I've tried worked. Realtemp doesn't work, everest doesn't work, speedfan doesn't work and I don't know what else may work. This is the other reason you should not buy from big computer companies.

[Hellsent 2.0]

Re: AD's water cooling experiment

Coretemp... And HW monitor

[signa]

Re: AD's water cooling experiment

Motherboard monitor *might* work.

http://majorgeeks.com/Motherboard_Monitor_5.3.7.0_Update_2_d5261.html

[=98=]

Re: AD's water cooling experiment

No, it sounds like an old motherboard or OEM one which never support temperature monitoring. They save costs by not installing temp diodes.

[AvalancheDiode]

Re: AD's water cooling experiment

I dismantled the testbed before further testing. There is no need for further testing. Water cooling is GOOD!! I am surprised by how little water can do so much cooling. The conclusion is that extreme overclocking still requires high water flow, but effective cooling for daily operation need no more than 100L/hr on CPU block.

While doing the test, CPU fan is still connected. The HP board won't boot without it. The fan, OTOH, runs at low speed and makes no noise during water cooling test compared to air cooling. The PSU had a lower load due to reduced fan speed and doesn't howl like dogs in mating season.

Once you go water, you don't want to go back. Now I want my rig to be water cooled. The challenge would be VGA water block. They are now priced too high, so DIY from sheet copper actually makes sense, provided if you have enough metal working knowledge, access to machining tools and plenty of time to spare.