How many radiators?

[plattman02]

hey guys im in a pickle.

so i ordered all my water cooling components at once. i got the xspc rx360 water cooling kit, but i wanted to mod it to cool my cpu and 2 x 560 ti's. i have everything i need and i was planning on having a single 140mm radiator somewhere in the loop along with the xspc 3x140mm. but somehow i ended up not actually ordering it and thought i did. is it worth buying that extra radiator and putting it in my system or will the xspc 3x140mm rad keep it plenty cool? i dont want to spend the money if its really not going to do much.

thanks!!!

 

[boiler1990]

Are you only cooling the GPUs? Have you read through the watercooling sticky at all?

There are some charts that will let you know what you should expect from your radiator in terms of heat dissipation, and you can find you components' heat using TDP as an estimate. Both of those will help you figure out if you need that second rad.

 

[plattman02]

cpu and gpus. and yes i have, but thanks!
also, should i have heatsinks on my graphics cards if i am using universal waterblocks?

 

[boiler1990]

Yes - VRAM will get pretty hot and the heatsinks will help get that heat off the chips themselves. You do need *some* airflow over them, but it doesn't have to be much.

Which uni blocks are you getting? I loove my MCW82.

 

[plattman02]

i also have two MCW82s haha.
but i cant find a 560 ti heatsink thats not sold out! it seems that swiftech stopped making them or something...is there another one i can get? if so, i havent been able to find it...

 

[boiler1990]

In reality you don't need one of those big uni-sinks. As long as you attach RAMsinks to each VRAM chip and the VRMs then you'll be fine.

I bought something like this since it was the cheapest at the time - just needed all the small heatsinks: http://www.frozencpu.com/products/12296/vid-162/Arctic_Cooling_VR005_Accelero_Xtreme_PLUS_Mounting_Kit_-_nVIDIA_GeForce_GTX_460.html?id=7fqPTRjv&mv_pc=243

 

[Lutfij]

just stumbled on these

http://www.frozencpu.com/cat/l3/g40/c21/s224/list/p1/Air_Cooling-VGA_HeatsinksCoolers-BGAVGA_Heatsinks-Page1.html

 

[rubix_1011]

I've used the same set of Zalman RAM sinks for years.

 

[noob2222]

I have my loop as follows. Pump/res > CPU > RX240 > 2x 6970 > Ex240 > res

What you want to do is cool the water down after the cpu heats it up. Try to avoid putting warm water into a dual gpu line as its already elevated and will only get hotter for the 2nd gpu. My temps are 38c max for the cpu, and 32c for the first gpu, 37 for the 2nd. My guestimation if I looped the gpus right after the cpu would be high 40C range and 50C range for the gpus.

Essentially putting a radiator between makes it the equivalent of a dual loop with only one pump.

 

[rubix_1011]

I have my loop as follows. Pump/res > CPU > RX240 > 2x 6970 > Ex240 > res

What you want to do is cool the water down after the cpu heats it up. Try to avoid putting warm water into a dual gpu line as its already elevated and will only get hotter for the 2nd gpu. My temps are 38c max for the cpu, and 32c for the first gpu, 37 for the 2nd. My guestimation if I looped the gpus right after the cpu would be high 40C range and 50C range for the gpus.

Essentially putting a radiator between makes it the equivalent of a dual loop with only one pump.

This is incorrect- you might see a 1-3C difference if you had the radiator as-is, or if you moved both radiators elsewhere in the loop. There is very, very minimal difference in temps when it comes to actual loop order.

Dual loops also aren't better than single loops- if you have a single loop and dual loop, both utilizing the exact same components, the single loop would perform better.

 

[noob2222]

This is incorrect- you might see a 1-3C difference if you had the radiator as-is, or if you moved both radiators elsewhere in the loop. There is very, very minimal difference in temps when it comes to actual loop order.

Dual loops also aren't better than single loops- if you have a single loop and dual loop, both utilizing the exact same components, the single loop would perform better.

so your saying components only add 1-3C to water temp? then how does the first gpu run 32C and the 2nd 37? thats at least 5 degrees. Say the cpu puts out as much heat as 1 gpu, then the first gpu would be 37C followed by 42c.

This is also on a highly overclocked system. At stock I might agree with this. I don't have a way to test it nor a desire to. I will know more when I get some temp sensors installed with a fan controller.

 

[rubix_1011]

I'm suggesting that the temp differences are less from being in succession and more to individual differences in how the blocks might be mounted/seated or even differences in the GPU components themselves.

Both of my GPUs (SLI GTX 560Ti's) run within 1C of each other, even under load. I actually would be alarmed if I had 2 watercooled GPUs in succession that were more than 3-4C apart at either idle or load temps. This would indicate either an anomaly of the GPU manufacture or an issue with the waterblock or how it is mounted. Watercooling has the ability to remove so much heat that is being dissipated through your blocks that there shouldn't be a difference of more than 5C at any single point in your loop (water temps) and the temps of identical hardware should be almost identical as well- in both idle and load conditions.

 

[boiler1990]

so your saying components only add 1-3C to water temp? then how does the first gpu run 32C and the 2nd 37? thats at least 5 degrees. Say the cpu puts out as much heat as 1 gpu, then the first gpu would be 37C followed by 42c.

This is also on a highly overclocked system. At stock I might agree with this. I don't have a way to test it nor a desire to. I will know more when I get some temp sensors installed with a fan controller.

The temperature reading you see from any sort of program like EVGA Precision or MSI Afterburner show you the chip temperature. However, the chip temperature is not the same as the water temperature.

 

[rubix_1011]

^Correct.

Still, both should be closer in reported temps than 5C+ difference.

 

[noob2222]

The temperature reading you see from any sort of program like EVGA Precision or MSI Afterburner show you the chip temperature. However, the chip temperature is not the same as the water temperature.

I never said they were, but they should be directly relational. Higher water temp = higher component temp. you can't cool a component lower than what the temperature of the water inside is.

your >5C difference is also highly dependend on one thing your not mentioning. water flow. The slower the water going through the components, the more heat is transferred (or dissapated). but seeing as my temps are still below 40c, I don't feel the need to rush out and buy a $95 pump. One way around it is ordering your components to be at their maximum efficiency.

Trust me, I know about heat transfer.

 

[boiler1990]

Of course they're related, but if we break it down...

The specific heat capacity of water is ~4.187 kJ/kgK (where a change in 1K = change in 1C).

Assuming you have 1 gallon of water in your loop (not unreasonable - I have about 9/10 of a gallon in mine), and said gallon weighs over 8lbs (3.6kg), that means you need:

Q = (3.6 kg)*(1 C)*(4.187 kJ/kgK) = 15.0732 kJ = 15073.2 W*sec

in order to cause the water temperature to increase by 1C. A 5C change requires over 75 kJ.

You're also dissipating most/all of the heat generated by the components from the radiators, so your net energy transfer (via heat) to the water will be essentially zero. Of course there are many other factors that play into it as well (like flow rate), but those are somewhat negligible unless you're talking very long amounts of time when these small differences accumulate.

 

[noob2222]

Of course they're related, but if we break it down...

The specific heat capacity of water is ~4.187 kJ/kgK (where a change in 1K = change in 1C).

Assuming you have 1 gallon of water in your loop (not unreasonable - I have about 9/10 of a gallon in mine), and said gallon weighs over 8lbs (3.6kg), that means you need:

Q = (3.6 kg)*(1 C)*(4.187 kJ/kgK) = 15.0732 kJ = 15073.2 W*sec

in order to cause the water temperature to increase by 1C. A 5C change requires over 75 kJ.

You're also dissipating most/all of the heat generated by the components from the radiators, so your net energy transfer (via heat) to the water will be essentially zero. Of course there are many other factors that play into it as well (like flow rate), but those are somewhat negligible unless you're talking very long amounts of time when these small differences accumulate.

That formula is right as far as I can tell, however, your cooling block only holds X water. Your formula is what's required to raise the entire loop, not just one tiny spot. When you examine that one spot, you have to take in flow, friction of the water moving through it, the surface area of said block, the material composition, and the heat source. Also consider the water is gouing from that spot, taking what it picked up there and carrying it to the next spot(gpu block). Repeat, and carry the heat to the third spot ... hence its a loop not a fishtank cooling solution. Also the warmer water tets, the less ability it has to carry more heat.

 

[rubix_1011]

As long as your loop has the ability to dissipate more heat than it is absorbing, reported temps should always stay relative to the working delta of your loop at equilibrium.

 

[Lutfij]

wonder where plattman02 is :/

 

[plattman02]

im just learning haha. thanks though guys all your input is highly appreciated! i actually did end up buying another 3 x 120mm rad, cause i found a nice deal on one today. im about to order those heatsinks right now too!

 

[Lutfij]

what'd ya pick up this time? how are you getting deals that fast?

 

[plattman02]

http://www.amazon.com/Radiator-3x120mm-30-FPI-Copper-nozzles/dp/B004YF2EGU/ref=sr_1_8?ie=UTF8&qid=1332992624&sr=8-8
i never even checked amazon before...its 30 fpi too

 

[Lutfij]

oh crap, that just means more louder fans