Watercooled setup produces higher temps than tuniq tower?

momo56

momo56

momo56

Soldato
Joined
18 Jul 2009
Posts
3,178
Right 1st of all this is my 1st setup so I apologise for any stupid mistakes I've made.

Basically the q6600 (B3) when watercooled maxes at about 65-66 degrees under load at stock frequencies. When i had the tuniq tower is was about 54 - 56 at load. The cpu is lapped and I'm sure I've applied the AS5 correctly.

My setup is basically the swiftech ultima 220 with an additional 120mm radiator, the EK waterblock for the 4870 x2 and the feser red UV fluid. The circuit is: pump>240mm radiator>cpu>120mm radiator>GPU>reservoir.

The 240 mm radiator is on the outside of the case with 2 fans. I've cut the side window to attach the 120mm radiator with the 120 mm radiator on the inside of the case. The case is just a zalman gt-1000 which I've just drilled a few holes into.

The 4870 x2 is about 30 at idle and doesn't go over 54 degress no matter how much I overclock the cpu.

So are there any simple ways to lower these temps? Anything I haven't thought of? I replaced the feser one solution with distilled water and the green stuff given with the 220 ultima which lowered temps about 2-3 degrees. Anything else?
 
sorry my specs are:

Zalman gt-1000 case
coolermaster modular 1000W
q6600 (B3)
asus blitz formula
OCZ platinum 5-5-5-15 1066mhz (2x2GB)
ati 4870 x2
raptor 300gb
Windows 7 32 bit
 
I#m no expert on these things, but isnt the loop order a bit wrong? You have your double rad cooling the water from the reservoir, then your cpu warming it up, the single rag cooling that, then it picking up heat again from the gpu, then going back into the res.

Wouldn't it going into the cpu, then double rad, then gpu, single rad, res work better? Or maybe even just cpu->gpu->double rad and get rid of the single one if the gpu isn't generating too much heat as you seem to suspect?

Again, I haven't done watercooling myself, so I could be wrong, but it just seemed a bit odd to me :)
 
I thought that the single radiator would cool the fluid as it has just been heated up by the cpu, and that a single radiator would be sufficient and that the double radiator would cool the fluid coming from the gpu. Normally under air cooling the gpu would reach temperatures of up to 120 degrees under load so I thought a double radiator would be sufficient.
 
I think you need to re organise your loop so it goes this way

res>pump>cpu>120mm rad>gpu>240mm rad>res

This way the water from the cpu is cooled before going into the gpu and then cooled again before re entering the res.
In your loop the hottest thing will be the gpu therefore it requires more cooling than the cpu.
 
Would there be any difference if warm water from the gpu entered the res? I mean it's still getting cooled by the double rad before it goes into the cpu anyway.
 
momo56 said:
Would there be any difference if warm water from the gpu entered the res? I mean it's still getting cooled by the double rad before it goes into the cpu anyway.
Click to expand...

Reading through the posts here and in other forums it is recommended to have the res feed the pump.

Therefore you want the water in the res to be as cool as possible as most pumps dont like it to be above 50c which it can be after heavy use of the systems graphics card i.e games like crysis etc
 
xman01 said:
it is prolly better if u
res pump rad cpu gpu res

if u have to use the 120mm rad then
res pump 2xrad cpu 1xrad gpu res
Click to expand...

Im no expert in watercooling however i have learned a lot from those here that have more elaborate loops than mine.

And the gerenal concensis is if you run just a cpu loop it should go like this

Res>Pump>CPU>Rad>Res

And if you run a loop with both cpu and gpu it should go like this

Res>Pump>CPU>GPU>Rad>Res

However as the OP is using 2 Rads i believe the best way for the loop to run is

Res>Pump>CPU>Rad 120mm>GPU>Rad 240mm>Res

this way the water is cooled sufficently before re-entering the pump. The pump the OP is using has a temp guide of 0 - 60 c and as the the Video card is a 4870X2 i can almost guarentee the water leaving it will be higer than 60c

The OP will find there is less chance of damaging his hardware and he should also lower the temp of his cpu as well.
 
The loop order doesn't make a whit of difference. It's counter-intuitive, I know! But it's true. After about 30 minutes the system equilibrates and the temperature is the same throughout the loop.

I think the whole premise of the question is wrong. I don't think you can compare a system that just cools the CPU with one that cools the CPU and GPU. The X2 has 2 CPUs dumping heat into the loop as well as the Q6600, so you're not comparing apples with apples.

That said, if the CPU is running as hot as that I would suggest the CPU block isn't bled-out correctly.

From memory the Ultima kit used to come with the Apogee GT block which wouldn't really be a good choice for a hot-running quad like the B3 Q6600, but more recently comes with an Apogee GTZ, which is very good.

Both these blocks are directional ie. they have an inlet and outlet side - make sure inlet pipe is connected to the lower, more central barb (marked IN) and the outlet is higher and (in the GTZ) offset. If you have mounted it upside down (outlet lower or sideways to the inlet) then you probably have air trapped in the block and, with the GTZ certainly, almost no water flow through the block as it's not designed to flow backwards. The GT is also directional, but less so. If you've had it apart then you masy have re-assembled it at 90-degres to the flow of the coolant, which doesn't work well.

That's the most common cause of issues in my experience.

And if your Apogee GT isn't fitted with the uprated O-ring (see under tweaking guide), then you want that too!
 
Last edited:
Of course the order of the loop makes a difference. Imagine you are a bit of water. You've just come out of the res and you're going through the pump. You go to the CPU and get very hot. You then go to the GPU and get even hotter, before getting cooled down again.

The GPU is just getting the hot water from the CPU, it's much more efficient to use the 120mm rad to cool water after it's been in the CPU.
 
I would actually suggest removing the 120mm rad and using the 240mm and getting a 360mm...its what I have done in my machine but using two loops instead of having it in one...see the links in my sig...and you will see what I mean...
 
Superewza said:
Imagine you are a bit of water. You've just come out of the res and you're going through the pump. You go to the CPU and get very hot. You then go to the GPU and get even hotter, before getting cooled down again.
Click to expand...

You'd think so, wouldn't you? But that's not really what happens. It's much more subtle than that. If it worked the way you suggest you would be able to feel a difference in the water temperature entering and leaving the radiator. And you can't. The actual temperature drop is only about 2C, and that's on a really big radiator.

Superewza said:
The GPU is just getting the hot water from the CPU, it's much more efficient to use the 120mm rad to cool water after it's been in the CPU.
Click to expand...

But if that was the case, then the GPU would be overheating, and it's not. It's the CPU that's too warm which suggests to me that there is poor heat transfer away from the cores, which is why I'm suggesting an air-lock.
 
WJA96 said:
And if your Apogee GT isn't fitted with the uprated O-ring (see under tweaking guide), then you want that too!
Click to expand...

I have a GT ordered from OCuk about a week after their cut-off date (Jan 15 2007) for including the 3mm o-ring but didn't get one included. Must have been old stock that my order come out of. Swiftech also says in that link "will be mailed on demand to US consumers free of charge", so not free for non-US then :( Oh well :)
 
+1 to what WJA is trying to tell you.
From my own personal experance I had a GT once that I installed 90 degress's off, but fully bled (leak tested flat then installed due to passive setup)
When I turned it round the right way my cpu temps droped 2c or 5% .
If you do have an airbubble inside the blocks pin matrix as well ...

And Bubo thermal images of his watercooling loop clearly show that the coolant is at a near constant temp throughout. But I was happy to beleive Cathar testing that's what, 10 years old now
 
Last edited:
WJA96 said:
The loop order doesn't make a whit of difference. It's counter-intuitive, I know! But it's true. After about 30 minutes the system equilibrates and the temperature is the same throughout the loop.
Click to expand...

100% Correct :)

Res->Pump is recommended to aid bleeding. What you do after this makes very little difference to individual and overall temperatures, providing you have sufficient flow. Flow is the key factor, a reasonable set-up has a flow rate around the 6 to 9lmp mark.

At 6lpm the liquid in 1l loop circulates every 10 seconds, most of that time is spent in the rad or the the tubing, a short while spent in the gpu block and a very short time spent in the cpu block. Measurement of the water in temperature and water out temperature of either block will show little difference, as a result loop order is, for all practical purposes, irrelevant.

.........
Quick experiment, mostly meaningless but it may help to put things in perspective.

Put 1l of water in a kettle, turn the kettle on for 10 seconds to simulate one full loop, give the water a good stir and measure how much the temperature has changed. Assuming the kettle is rated at 3,000W and a PC CPU + GPU combo is around 300w, divide the measured temperature difference by 10.

Answer: (10 x 3000) / (4200 x1) / 10 = about 0.7C. This is without any form of cooling.
.........

WJA96 said:
I think the whole premise of the question is wrong. I don't think you can compare a system that just cools the CPU with one that cools the CPU and GPU. The X2 has 2 CPUs dumping heat into the loop as well as the Q6600, so you're not comparing apples with apples.
Click to expand...

:D
 
Loop order makes little to no difference to temps, the only important thing to do is keep a head of water above the pump / like the res.

What fans (and voltage) are you using and are you blowing/sucking cold air over them or air thats been though the case first ?
 
momo56 said:
The 4870 x2 is about 30 at idle and doesn't go over 54 degress no matter how much I overclock the cpu.
Click to expand...

As you stated under air this card would hit 120c and now its max 54, that would answer why your load temps on your Cpu seem nothing special against air cooling..a lot of heat is being taking away from the Gpu, which is good but means that ur cpu cooling is being effected. Thats why ur seeing the cpu temps you are.

To get better overall temps you need either a 360 and that 240 like someone allready suggested or to make it into 2 loops.
 
The reason why the temperture doesn't vary that much between point to point, e.g. either side of a cpu, is that its not stictly speaking "units of temperature" changing hands when water goes through a heat generating component, its rather units of energy. The final temperature water will get to is an equilibrium condition which depends on the ambient temperature are the rate at which the radiator can extract energy from the water. It may be a fair number of joules of energy passing from the cpu to the water but as far as a discrete in/out temperature change that is dependant various properties such as the specific heat capicty of the fluid as well as flowrate, and does not automatically mean that because the cpu is chucking out X watts that the water temperature has to vary by a large amount. With the flowrates we are talking about there simply isn't time for the water to change in temperature that much across a cpu/gpu/whatever. There is a temperature difference across them of course, its just that it is not as big as you would expect, but is simply obeying the laws of thermodynamics.
 
You must log in or register to reply here.
Back
Top Bottom