DIY Radbox

Soldato
Joined
3 Nov 2004
Posts
9,871
Location
UK
Got a little bored over Christmas, thought I'd make an external radbox. Haven't used water in a couple of years. Just wanted to take advantage of the cold sink I have on my windowsill. Designed to sit behind the curtains on the side of the house that never sees any Sun. Self powered, just two rpm sensors back to the PC.

Components.

D-TEK FuZion CPU Block + Nozzle Kit
Laing DDC3.2 18w Pump With PetrasTech Top
ThermoChill PA120.3
Masterkleer 7/16" Tubing
Swiftech MCRES-Micro
3 Yate Loon D12SL-12
3 Sanyo Denki 1011
MEANWELL SMPS 12V PSU
G11 Coolant + PT_Nuke

Oh and a B&Q toolbox.

Got the PSU cheap on Ebay, overvolts to 13.6V. Made a PWN controller for the inline thermistor probe. Switchable from resistors for 12v/7v/5v to variable PWM. Cost about £10 to make. Switches the high flow San Ace off at cpu idle, then increase rpm as the loop heats up. The Meanwell psu switches by relay from the main PC.



 


Replaces a Thermaltake Ultra 120, lapped with two fans in push/pull. So just about as good as air gets. Using Prime95 small fft's

Ambient 21C. Air cooled 30C/60C (65C after an hour or two) idle/load on a Q6600 @ 3.6GHz 1.44v.



Closed room, ambient 25C. Water cooled 26C/44C (Stabilises at 47C after an hour or two) idle/load on a Q6600 @ 3.6GHz 1.44v.



Window open with outside air 7C, room ambient 21C. Water cooled @ 3.6GHz 1.44v. 28C load
icon14.gif




Window open with outside air 1C, room ambient 12C. Water cooled @ 3.8GHz 1.48v. 28C load.



Window open with outside air 3C, room ambient 15C. Water cooled @ 4.05GHz 1.60v. 38C load. Over 250W from the CPU.



Closed room, ambient 26C. Water cooled, temps stabilised at 51C full prime ffts on a QX9650 @ 4GHz 1.38v.



http://valid.x86-secret.com/show_oc.php?id=306760
 
Last edited:
oooooooooooooo nice work! so what you gona do when it comes to the summer?

Its priming for 24/7 settings at the moment. Typical use, window's shut and the door's open. Central heating's on, rooms 21C. 3.8GHz @ 1.48 loads around low 50's.



Given the same conditions the water is around a 15C improvement from the best air I could achieve. The point of the cold air examples apart from a laugh, is that I can make use of cold air without cooling the whole room. Just watch out for condensation on the pipes in the warmer room.

Its worth noting that actual water temperature isn't that much higher than the ambient once the loop reaches equilibrium.

e.g. Ambient 23C, water 26.6C, cpu 42C, core 56C. Flow rate is in excess of 2.0GPM





Final CPU case (not core) temperature.

Ambient + system load (inc pump) x radiator C/W + CPU Load x block C/W + CPU load x TIM C/W

23 + 224*0.018 + 206*0.07 + 206*0.001

23 + 4 + 14.4 + 0.2 = 41.6C

Thats for the IHS temp, one rough way of estimating the core is to double block on the cpu.

23 + 4 + 206*2*0.07 + 0.2 = 56C
 
Last edited:
Very nice, looks good and is obviously giving great temps, give yourself a pat on the back :)

I'm designing one soon, which will be going into mass production.
 
I'm designing one soon, which will be going into mass production.

Really? Can you keep me updated as I'm looking for something like a Zalman but with temperature controllable in a smarter way using fans if required.
 
Window open with outside air 3C, room ambient 15C. Water cooled @ 4.05GHz 1.60v. 38C load. Over 250W from the CPU.

so this means u can push ** cpu further as its no where near 70 c ?
 
Back
Top Bottom