jimbearsfan1

Hello...new to the forum. Nice thread, I'm a user and supporter of Coolit products for quite a while now.

I am doing a case with a Boreas unit and a FZE, with each designated to specific cooling assignments.

I'm using the Boreas to cool my tri-SLI setup (EVGA GTX 280 FTW x 3), chipset and mosfet blocks (Koolance NB/SB full with mosfet blocks for my Classified).

I'm using the FZE strictly to cool the CPU (i7-975)

I am using stock Coolit products....tubing, coolant, blocks, fans, pumps, for both units. Wanted to do a dual Boreas setup, but couldn't make it fit.

First question, do you feel this is the proper distribution of cooling units? I thought about doing it the other way around, but I didn't think that the FZE would be enough for the 3 cards and the chipset blocks.

I plan on overclocking the CPU, moderately (around 25%, so in the 4.1 to 4.2 range), and the graphics cards come overclocked from the factory, so I don't plan on pushing those any further.

Will the FZE do a good job on the CPU overclocked as I've described above? I currently have a FZE cooling my QX6700, and it's overclocked from 2.66 to 3.33, with my idle temps being in the 29c range (ambient in my room is typically in the 25c area), and load temps being in the low to mid 40c range.

Next, is there anything to be gained by modifying the Coolit products....changing pumps, cpu block, coolant, etc.? From what I've read in the thread it seems there is performance to be gained with modification.

What are your thoughts on adding a rad and res? Is there an advantage to this, and can they be shared with both Coolit units?

Just a few thoughts...appreciate your input.

Regards.

TiGa

The best setup I could suggest in your case is the Boreas on the CPU to have the water stay sub-ambient and a FZE + rad on the GPU.
A single FZE is really not enough for an i7, you would get similar temps as using a 360 rad alone. Running the GPUs with sub-ambient water doesn't give any extra advantage.

A Boreas would give you temps on an i7 comparable to what you're getting now on a quad with a FZE. i7 chips run really hot, especially when overclocked.

I highly recommend changing the CPU block, an Apogee GTZ or Heatkiller will do a much better job than the CoolIT block. An after-market block is more efficient at extracting heat from the CPU.
The CoolIT pump is a bit noisy and not powerful enough once you change the CPU block for a more restrictive one. I would recommend a MCP350 or MCP355. I had better results with them because of the higher head pressure compared to the more expensive MCP655.

It is better to keep your loops separated. If you use only a single loop, you will see your CPU temps raise a few degrees when using the GPUs for rendering or whatever, even if the CPU is idle.

A single FZE with a rad will perform better but you won't be able to go sub-ambient as the rad will heat up the water instead of cooling it.
A res is another place where sub-ambient water will heat up but you just need to insulate it. The advantages outweigh the disadvantages. I'm using the MCP335 with the XSPC res top.

I hope this helps you a bit.

TiGa

__________________
CoolIT TEC Club

jimbearsfan1

Thanks TiGa. A couple of more questions.

Opinions on other fluids such as the Nano Fluid?

Is it best just to leave the res out of the picture completely?

On the pumps you mention, is the flow adjustable, and does it need to be turned down to work properly with TEC cooling?

Assume you recommend sticking with the 1/4" Norpreen tubing, and using 1/4 fittings for the cpu blocks you mentioned?

If the rad will actually raise the temp, why use it at all? Not clear on that idea.

Appreciate the help very much.

Regards.

ChielScape

id just make it one big loop. a single MCP355 with restop, and 3/8" fittings.
the freezone has 2 blocks with 3 TECs, the Boreas each have 4 identical blocks, which means you're totalling 10 blocks... i recommend running 5 and 5 in parallel.
(3/8" splits into 2x 1/4" -> each 1/4" passes trough 1 boreas and half a FZE, then Y back into 3/8")

id run it like this, res/pump -> CPU block -> GPU 1, 2, 3 -> Y-fitting -> 2 sets of chillers -> Y-fitting -> back to res/pump.

recommend you use nickel plated blocks, like koolance makes. also recommend koolance or FluidXP+ Extreme coolant (the Extreme has a certain anti-corrosive additive) to help against the corrosion that would occur if you mix the aluminum chillers in a loop with copper blocks.

TiGa

I'm still using CoolIT's own coolant to minimize the chances of reaction between the anodized aluminium chiller blocks and the copper CPU waterblock.

The res is useful for bleeding out air bubbles and be sure that there is enough coolant, which can be hard to see with opaque tubing.

I'm using Koolance 1/4" ID 3/8" OD compression fittings with CoolIT's 1/4" Norprene. Just like with any watercooling loop, high flow is always better. The MCP355 pump is not adjustable but since more flow is always better, the adjustable setting is kind of pointless. It takes a good pump with high head pressure to be able to push enough water through 1/4" tubing.

There is a difference depending whether you have sub-ambient water or not. If your water is over ambient, the chiller behaves just like a rad, removing heat from the water. A radiator will never be able to take the water below ambient temperature. The moment the water becomes sub-ambient, the rad can only heat it up instead of cooling it down.

It's all relative to the temperature of the water like for example, blowing on hot soup to cool it and blowing on cold hands to warm them.

A single stock FZE on the CPU or GPUs would be barely enough to take the water sub-ambient at idle. So adding a rad could hurt temps a little a idle but give a big help at load. The CPU will appreciate sub-ambient water at full-load much more than the GPUs.

TiGa

__________________
CoolIT TEC Club

jimbearsfan1

Thanks again for the input.

So, would the idea of one big loop using both the Boreas and the FZE in series, with the Heatkiller 3.0, rad (double?), 355 pump with res work well?

My thought is: res/pump -> cpu -> mosfet -> chipset -> gpu 1-2-3 -> rad -> FZE -> Boreas -> back to res/pump.

My thinking is that the fluid hits the CPU at it's coldest this way, flows through the blocks (all blocks are Koolance by the way), goes to the rad to start the cooling process, then the FZE to further chill, and finally the Boreas to really drop the temps.

I also had read that for the TEC cooling to function properly, the pump flow needed to be lower to give the TEC's time to cool the liquid as it passed through, which was why I mentioned lowering the flow on the pump.

If the one big loop idea isn't the best choice, perhaps putting the CPU and chipsets on the Boreas, and the GPU's only on the FZE might be a better choice? I could use the same flow as above, only use the rad and 355 pump with the Boreas only, and leave the FZE stock with the GPU's.

Hope these questions don't seem too fundamental, just want to talk to some experts to get the setup right the first time.

Thanks again TiGa, and ChielScape for your recommendations.

Regards,
JBF

ChielScape

i would ditch the rad unless your water temps go over ambient.

this is a myth

TiGa

I've got a temp sensor on the first and fourth chiller blocks of my dual FZE setup and the temp difference is about 0.7C, even 3 GPUs in series will show at most a 1C difference so you can expect at worst a 2C differential between the hottest and coldest point of a huge loop.

Now with a huge loop like that, you would probably need two MCP355 pumps in series. The lack of flow is mainly caused by the 1/4" tubing. To be close to their peak efficiency, CPU blocks need around 1.0-1.5GPM. With two MCP355 pumps and a CPU-only loop, I have barely 1GPM. With the chipset included in the CPU loop, I had 0.6GPM and CPU temps 5C higher.

The Heatkiller is a less restrictive block than the one I'm using right now so you might get around 1.5GPM with CPU-only or 1.0GPM with the NB/SB and MOSFETs included with the CPU.
These would be very good conditions to run a CPU loop with or without MB blocks but add 2 other chiller blocks, you get less flow. Add 3 GPUs, you get even less flow.
That's why I was proposing separate loops. The CPU loop would be able to keep its water sub-ambient so no rad would be needed.

Adding a rad usually makes it a lot harder to go sub-ambient but I'm not sure even the Boreas alone could keep its water sub-ambient at with those GPUs at load. As I mentioned earlier, a rad will heat up sub-ambient water but cool down over ambient water. It's all relative.

Anyways, half the fun of modding is trying things out and tinkering, when time allows it, of course. You can start with FZE + rad on the CPU and Boreas on the GPUs (or the contrary) and then do a switcheroo. Once you've got enough extra fittings and tubing to play around with, trying every possible setup you can think of is only expensive in time but not money.

If it was me (I'm not you), I would put all the chiller hardware on the CPU+MB and get a 360 or 480 rad for the GPUs. Actually, that's what I'm doing right now.
I got two GTX295s on two 240 rads. I got the chipset on its own loop with a MCR120 and they all seem very happy.

TiGa

__________________
CoolIT TEC Club

XenoMopH

Since most of the things are already said, I'd say to just go trial and error.
Try some various settings, blocks and combinations. The thing about low flow is better for TEC's is a myth. That's what I thought for a while, until I bought myself a MCP350 with an xspc topreservoir. The shorter the coolant is in a res, the better, so the res.top is perfect It sure took down my load temps. And the addition of a HK3.0 took off yet another few C's.

__________________
CoolIT TEC Club
PII 940BE Stepping: CACVC AC 0851 APMW @ 4140MHz
Add your Phenom 2 cpu-z validation HERE
System: Sulaco PICS
9850 OC @: 3.31GHz

jimbearsfan1

All..

Thanks. Great info and opinions. I have learned much from your comments, and am going to incorporate as much as I can into the build now...while it's still in the build stages.

I appreciate all your help, and will soon post some pics as I move things along.

I think I'm going to use the Fluid XP Nano Fluid, seems to have all the requirements mentioned regarding reaction, and is also non-conductive.

Thanks again.

JBF