WTK: One Rad good enough or two

[TaranScorp]

I can fit a RX240 Rad in the bottom of my case running two fans in push configuration. I am running a 3770K at 4.4 with a Archon cooling the chip. Linx get's the chip to 82c load and my encoding program gets the chip to 72c load.
for grins I want to try 4.6 or 4.8 but want to switch to water first. I also found out I can fit a EX140 rad in the back in push/pull.
The question is will the extra EX140 on the back make any noticeable difference in cooling?

 

[BrightCandle]

It looks like an 3770k gains about 60% power consumption from being overclocked at 1.35V @4.7. So given that and its TDP of 77W we can estimate consumption at peak around 125W. That is half the wattage of a 3930k or a 9x0 chip, which is amazingly low. Equally it means that for watercooling a single 120mm thick radiator will achieve 10C water delta at 1000rpm's or so. Simply put I doubt Ivy Bridge 4 cores benefit much from water cooling when overclocked.

Nevertheless 2x120mm radiators will give you around 5C delta, which is a very good amount of cooling. Adding the additional back radiator will decrease to about 3C, so you might see up to a 2C reduction on the CPUs peak temperature for the additional rad. Alternatively you can drop the fans down to 800rpm and likely see around 4C delta and have complete silence.

I wouldn't bother with the additional radiator, this chip simply doesn't put out enough heat to need it.

 

[TaranScorp]

Wow, after the 3770K came out for sale I've read hundreds of posts of people complaining about the heat the 3770K gives off. There are threads galore of people doing all kinds of things to bring the heat down, Sanding,delidding, changing the tim in the cpu, peltier
If any of those people read this they would probablty pass out

There's also 4.8 or 4.9?

 

[BrightCandle]

The Ivy Bridge chips get very hot, but that is because the thermal interface between the silicon and the heatspreader is of poor quality compared to what Sandy Bridge uses. It may well be cheaper for Intel but its impact on thermal transfer is quite problematic.

But none of this has anything to do with watercooling. Heat output from the top of the CPU into the water and then into the air is all we can control at this point. In order to cool the CPU as best as possible we need to ensure we have sufficient exchange with the air such that the water is kept relatively cool, within say 10C of the ambient air temperature.

Watercooling wont fix the flaw in Ivy Bridge's attachment material, the CPU will still run hotter than it would with a better material. In the same way that thermal paste from the core to the spreader increases the temperature so will the water between the radiators and the air. All these materials between the actual silicon and the eventual exchange with air decrease the quality of the cooling. Watercooling makes up for this by ensuring their is more than enough exchange with the air to undo its impact and reverse the trend such that at load it performs better than air cooling direct on the CPU can.

My concern with speccing your water cooling is to ensure it will meet the thermal output and do so in such a way that you will see benefit from the water compared to high end air. I can't fix the problem with the CPU and the heat resistor Intel has put in underneath the spreader.

 

[TaranScorp]

Very informative and helpful, Thank You.
What do you think of the idea of chaging the Tim in the chip to aid in cooling?

 

[_Rick_]

A Peltier element would probably help though.
Then you can get the full benefit from watercooling, because you'll be producing more heat, but you'll get a larger delta-theta at the bottleneck, allowing for more energy transfer.
Probably a better idea than dropping an extra rad. Though that one might become necessary, depending on how hard you drive the Peltier.

 

[SenorRed]

The way my setup is, the water passes through a large 3x120 rad, then through both of my video cards, through a single 120, and then to the 3770k. I would say based on my setup, a single 120 isn't enough, but a 240 would for sure do the trick. My load temps at 4.6 are in the high 50's low 60's, So I can only imagine they would be much improved with a single loop dedicated to just that.

 

[TaranScorp]

Yea, this water cooling stuff is interesting, my first attempt.
I found out I can fit a thin(35.5mm) 240 rad at the top of my case and a thick(58.5mm)240 at the bottom of my case. I think that would put the fear of WC into that 3770K Or is it way overkill? So far it sounds like the one thicker 240 rad should be good enough for just a 3770K cooler.

 

[BrightCandle]

Lets put it this way - with a 120mm fan thick radiator you are able to cool the water such that it was 10C above ambient at full load, and it'll take a while to reach that temperature.Given an infinite number of radiators (and an infinitely good pump) you would only be able to improve the CPU temperature by 10C from that point, by reducing the water temperature to 0C above ambient.

The temperature of the CPU is determined by the water temperature and the quality of the waterblock, so all things being equal the cooler the water is the cooler the CPU is. 5C water delta is already a seriously good cooling setup.

Watercooling is surprisingly simple - you just need to balance the heat in to the heat exchange out of the loop. To do that you just add up all the power of the devices your cooling and the cooling potential of the radiators for a given water temperature. That really is all there is to it. If you put 240 thin and 240 thick radiators in there you really need to put a waterblock on your GPU to make that worthwhile.

I run 2x 360mm worth of radiators and I am marginal on cooling (about 12.5C above ambient with everything fully loaded, which never happens thankfully). But then I have 2 7970's on that loop, plus the motherboards VRMs and MB NB combined with the CPU.

 

[Smoblikat]

Wow, after the 3770K came out for sale I've read hundreds of posts of people complaining about the heat the 3770K gives off. There are threads galore of people doing all kinds of things to bring the heat down, Sanding,delidding, changing the tim in the cpu, peltier
If any of those people read this they would probablty pass out

There's also 4.8 or 4.9?

Ivy bridge still has high temperatures, it just doesnt have high heat. You assume those two words are synonomous (like most people).

 

[TaranScorp]

I have to do some ciphering on the heat temperture thing. But right now with no research done it's a mind bender.
I only have a iGPU to go with my i7-3770K and don't think it needs cooling.

 

[BrightCandle]

I have to do some ciphering on the heat temperture thing. But right now with no research done it's a mind bender.
I only have a iGPU to go with my i7-3770K and don't think it needs cooling.

I don't know about you but at school in Physics we did this calculation for the loss of heat from a house. We had various details of the materials we could put in the walls and how much heat they transferred and we had a difference in temperature from the inside to the outside. Different materials between the air in the house and the air outside the house transferred or more less heat, which affected how much heating was needed for the house to maintain its temperature or alternatively how long it would take to cool to near ambient. It was a basic model but the principles from that still apply here.

So there two things going on, there is a so called volume of heat and there is the actual temperature. They are tied together by how much cooling is applied. The amount of energy transferred is dependent on the difference of the temperature of the two items and the amount of resistance there is to the heat transferring between the two materials. A higher difference in temperature produces more heat transfer, a material with less thermal resistance also increases heat transfer. So if the material gets worse the temperature goes up which increases the difference in temperatures and balances such that the same amount of energy is transferred.

Practically what this means is if we had the perfect material to cool the CPU then its temperature would be the same as the ambient air. Metal however isn't perfect and thermal paste is a long way from perfect and so a temperature gradient is caused in each material such that he temperature on oneside is higher than the other to compensate for the poor ability of the material to transfer heat. Because there is now a higher difference in the temperatures there is more flow of energy which now evens out.

I don't know whether that helps but simply put we can't do much about the materials and their impact on absolute temperature. What we can however do is ensure we have sufficient cooling for the amount of heat produced and the temperature will then be governed by the quality of the materials. Alas with Ivy Bridge the materials are quite bad, much worse than SB and previous CPUs of recent years and its this thermal resistance which is causing the high thermal differences to ambient we are seeing. You can base your build on 10C of water delta cooling and you'll be just 10C away from the absolutely best water cooling possible. Adding more radiators from that point can never achieve better than 10C less no matter what you do, physics is kicking you in the butt.

 

[OVerLoRDI]

Yea, this water cooling stuff is interesting, my first attempt.
I found out I can fit a thin(35.5mm) 240 rad at the top of my case and a thick(58.5mm)240 at the bottom of my case. I think that would put the fear of WC into that 3770K Or is it way overkill? So far it sounds like the one thicker 240 rad should be good enough for just a 3770K cooler.

two 240 rads for a ivy bridge is overkill. I say add the second rad AND your GPU to the loop

 

[_Rick_]

http://www.tetech.com/Peltier-Thermoelectric-Cooler-Modules/High-Performance.html

23.20 for a 30x30mm peltier element.
If you run it at 12V ~3A you will have a significant drop in temperatures on the cold side, and around 40W (12x3 = 36 ) of extra thermal output.
Considering the cost of an extra radiator, this looks like a pretty sweet deal.
There should be no cold bug issues, as temperatures should stay above 220 deg K, even in idle.

Only challenge would be to mod the heatsink mount, and get an ideal thermal interface between CPU/peltier and peltier/block.

I'm also really interested in seeing how this would turn out

 

[TaranScorp]

_Rick_ , Going to Water Cooling is a step for me from air cooling for the last 20 years, peltier is way over my head for now. In the last year I just switched to sata drives and in the last month bought a SSD for the operating sytem. Only been building my own comps in the last two years.
I just don't like throwing things away if they still work.
I'm typing this on a Dell 4400 P4 1.7GHz 1GB ram Win98se

 

[_Rick_]

I'd argue, that putting more or less conductive liquids in flexible tubing in your compter poses a much higher risk than dabbling with a cheap peltier plate

And, again, especially for Ivy Bridge, the water cooling alone will have minimal impact.
So - if you want to enhance your setup, a Peltier would be an effective investment. An additional radiator (especially inside the enclosure) not so much.
Radiator placement alone (getting it out of the box, so you don't dump heat inthe case, or use warm air from the case to cool the radiator) is probably going to be a more effective measure, as you can decrease radiator temperatures to a lower baseline.

 

[BrightCandle]

Peltiers aren't exactly a bed of roses. For every watt they transfer they consume a watt of power. So an overclocked IB is going to need something like 130W of peltier, which is not small. The below ambient temperatures cause condensation and to top it off they require pretty special mounting. If it fails, which they do quite readily, they turn into resistors of heat and your machine goes pop.

Very few people use them today because its almost impossible to get one small and efficient enough to deal with the heat density of a modern PC, and the benefits aren't worth the drawbacks. Just take a look at the exact same discussions on the xtremesystems forums and you'll see this come up a lot and the experts there that have used them will guide you to either water or phase change depending on how extreme you want to go.

 

[TaranScorp]

Yea, I was originally going to mount the rad on the outside of the case but got caught up in reading how people like to keep everything inside the case nice and neat.
The kits I am looking into come with brackets to mount the rad at the back of the case.
So I'm going to buy the kit with the RX360 Radiator since it's only $20 more then the RX240 kit and mount the rad outsde in the back.
My Storm Trooper case just happens to have two holes cutout in the back with nice rubber grommets for water cooling tubing
Thanks everybody for the help.

Petier is out of my league.

 

[BrightCandle]

Let me know how you get on, I would like to see how much improvement you see on Ivy Bridge with a well specced water loop.

 

[TaranScorp]

okeee dokeee
Not until I get paid next Friday. Then I'll order.

 

[Interitus]

FWIW, I use a 3x120 on my i7 920 and it's more than enough. I would imagine even a budget 240 rad like the Swiftech MCR would be plenty for what you're trying to do. At 4ghz on the 920 with an MCR320, I keep my Gentle Typhoons at around 1k rpm or lower depending on ambient temp.

With ambient about 80F, I idle at 33C, LinX loads up to 60'ish. And this chip runs way hotter than an Ivy would I'd imagine. I am on dual DDC pumps though.

-edit-

Forgot to add that my 920 idles pretty hot as well, so the idle temp is probably much higher than normal. On air on a Venomous-X HSF, it would idle in the high 40's even at stock, but didn't creep up that much when loaded. It's an odd, hot chip, so keep that in mind too.