FAO: Westie (Watercooling)

Toff

Toff

Toff

Associate
Joined
11 May 2006
Posts
46
Location
GX Bucks
Hello Westie,

Just looking through the thread where everybody have posted pics of their cases. I was really interested when I saw the picture of your water cooling system.

I noticed that you had a three-way splitter - I thought the water had to flow from device to device to device (etc) with intercooling in between to work effectively :confused: (but then again, I don't know a great deal about it!)

Is there any chance you could post a diagram / schematic of how your watercooling is set-up? (I can't make out where some of the pipes are going)

Cheers - Toff.

Westie-DSC000480.JPG
 
also the key to watercooling is to have as little restriction as you possibly can for the components you are using... exesive use of joints is bad news, as is exsesive tubing. the idea is to use as little tubing as humanly possible :)
 
His water is flowing out of his res and into his pump, then getting pumped up to his radiator to get cooled before flowing down to his CPU block, as the water exits his CPU block it passes through a Y-splitter where it is being split into two loops, one loop going to his NB waterblock and the other to his his gfx card block, the two loops then rejoin at another Y-splitter before the water is sent back to his res.

Just an educated guess of course (although i'll put money on it being bang on, lol)

:)
 
Thanks for your replies.

Bundles - that makes sense, good detective work :)

I guess by splitting into seperate loops, you're stopping the water being heated more and more by each component it goes through (i.e. when in series) ?

How important is it to cool the chipset ? (IC7-MAx3 has a heatsink with fan - which I've removed the pad from and carefully replaced with silver based thermo).

- Toff
 
not needed at all...
only advice i would give is if you have a passive cooled chipset... make sure you replce the airflow you would lose by takign away the CPU fan / GFX fan.


your comments about stopping the water being heated more and more is erm,... well wrong.
the whole point of watercooling is to get as much HEAT into the WATER as humanly piossible this is a fine balance between

tubing length ( restriction )
pump head( or pressure - how far tit can pump water up a vertical hose)
pump flow rate ( NOT as important as pump head)
block restrictiveness... usualy measured in loss of head
c/w (pretty much universal idea of how well a waterblock performs - lower the better )
 
jidh007 said:
What's an ideal flow head to use? I'm having to have my radiator out of my case, probably an extra 25cm of tubing will that affect much?
Click to expand...

About 21 feet is supposed to be ideal.

No, an extra 25cm of tubing won't hurt much at all.

Here are the results of the testing I did with my components, adding each one and measuring the flowrate. This was done with two AQX50Z in series @ 13.8v, so about 23 feet of head:

system_flowrate.jpg


In my testing, I played with parallel loops but found that parallel can be VERY bad - the water will take the path of least resistance, leaving the more restrictive portion starved of water. I was surprised how serious the effect was.
 
Using y junctions add resistance as the barbs narrow the channel. Think this was one reason why dandgerden brought out the tdx block and why the tdx is better than the rdx block.
 
Agreed, I considered running components in parellel, but after a few tests I found that water was taking the path of least resistance, effectively ignoring the GPU, and cooling the Nbridge more. Not good. This may just be my noob-ness and i've missed somthing, but it's what happened to me *shrug*

I think the best idea would be to run 2 loops seperately...

Still, the setup looks awesome :D, what case is that?
 
R B CUSTOMS said:
not needed at all...
only advice i would give is if you have a passive cooled chipset... make sure you replace the airflow you would lose by taking away the CPU fan / GFX fan.
Click to expand...

Thanks, I'll point some fans at it :)

R B CUSTOMS said:
your comments about stopping the water being heated more and more is erm,... well wrong.
the whole point of watercooling is to get as much HEAT into the WATER as humanly possible this is a fine balance between
Click to expand...

Understood.. but if the water passes in series from CPU to VGA to Chipset to MEM (for example) isn't the water going to be a lot hotter when it reaches the MEM when compared to a system where you split the tubing into separate parallel tubes for each component that needs cooling?

or is it important to keep things in series to maintain pressure / flow :confused:

Thanks - Toff.
 
Toff said:
Understood.. but if the water passes in series from CPU to VGA to Chipset to MEM (for example) isn't the water going to be a lot hotter when it reaches the MEM when compared to a system where you split the tubing into separate parallel tubes for each component that needs cooling?
Click to expand...

No. Delta T (water@CPU / water @ mem) is likely to be 1'C at most.
 
Jimbo Mahoney said:
No. Delta T (water@CPU / water @ mem) is likely to be 1'C at most.
Click to expand...

Thanks, you've saved me a fortune in Y-splitters, clips and unnecessary leak points :) As I've been given a free chipset water block, I may as well stick it in there (I'm aiming for a v.quiet PC).

I now plan to plumb-in like this: Pump+reservoir > CPU > Graphics > Chipset > radiator > (back to pump+reservoir)

Is that right, or should the radiator go between the Pump+reservoir and the CPU? Is it necessary to have a vertical pipe open to the atmosphere (as seen on a few people's installs?) and if so, where abouts in the line should it go?

Thanks again - Toff.
 
Order doesn't matter as long as the pump has a nice clean low-restriction feed (ie its intake side) - reservoir is great for this.

I don't think systems have open-air tubes - they are normally bleed / fill tubes but if you have a res, you won't need a bleed tube.
 
Jimbo Mahoney said:
I don't think systems have open-air tubes - they are normally bleed / fill tubes but if you have a res, you won't need a bleed tube.
Click to expand...

Thanks. Do you mean: if the Reservoir is at the highest point (vertically), I'm unlikely to get bubbles?

I like the hint somebody made about not adding the additive until later - to avoid foaming!

- Toff.
 
Mikebert4 said:
Agreed, I considered running components in parellel, but after a few tests I found that water was taking the path of least resistance, effectively ignoring the GPU, and cooling the Nbridge more. Not good. This may just be my noob-ness and i've missed somthing, but it's what happened to me *shrug*

I think the best idea would be to run 2 loops seperately... QUOTE]

I run my system in parallel (dual CPU) and both CPU's run at the same temp even though i thought the lower one would get all the flow, going to repipe it soon with 15mm tube splitting off to 2x 10mm to each cpu instead of all 10mm and then split again to the GPU and Chipset when I get a block for that.
Did it in copper to try and make it look neater!

Everest report:-
Sensor Properties:
Sensor Type Winbond W83792D (SMBus 2Fh)
GPU Sensor Type Winbond W83L785R (ATI-I2C 2Dh)
Motherboard / Video Adapter Asus NCCH-DL / Abit vGuru
Chassis Intrusion Detected Yes

Temperatures:
Motherboard 38 °C (100 °F)
CPU1 41 °C (106 °F)
CPU2 41 °C (106 °F)
GPU 33 °C (91 °F)
GPU Memory 32 °C (90 °F)
SAMSUNG SP2504C 22 °C (72 °F)

EDIT that is at 100% load running SETI not idle lol

Piccy
b162c3d9.jpg
Click to expand...
 
Last edited:
h2ocooler said:
I run my system in parallel (dual CPU) and both CPU's run at the same temp even though i thought the lower one would get all the flow, going to repipe it soon with 15mm tube splitting off to 2x 10mm to each cpu instead of all 10mm and then split again to the GPU and Chipset when I get a block for that.
Did it in copper to try and make it look neater!
Click to expand...

Both CPU 'loops' have the same resistance (minus the few cm of copper tubing in the top, which is practically negligible resistance).

For anyone else thinking of going parallel, ONLY do it if each branch has identical components.

Nice plumbing btw! ;)
 
Jimbo Mahoney said:
Both CPU 'loops' have the same resistance (minus the few cm of copper tubing in the top, which is practically negligible resistance).

For anyone else thinking of going parallel, ONLY do it if each branch has identical components.

Nice plumbing btw! ;)
Click to expand...

LOL thanks, should be its my Trade!
 
You must log in or register to reply here.
Back
Top Bottom