A nervous request for opinions on WC setup.

[Trekari]

Hello to all who read this:

My name is Jason and I'm new here. I post ...rarely over at bit-tech.net and pcper.com and have even visited overclockers.com before and that is where I first ran into BillA's meticulous testing methods.

I'm a bit nervous to post here because I've seen the ease in which threads can get very technical and..harsh, but frankly I feel there is no better place to go for anyone who has serious interest in watercooling and getting the best setup.

Anyways, my friend gave me his Koolance 601 setup..and I can't stand it. I've built my own custom Lian-Li PC-60 watercooling setup with a Maze 3, BIX, and Via Aqua 1300 (of which my opinion is that pump sucks), but recently sent that off to my poor dad who lives in Alabama and was losing hearing from his noisy PC fans.

The Koolance is mounted in this desk seen here. My version has the enclosure on the left side and the file cabinet on the right, but that is irrelevant. The PC is stored inside the enclosure, and although I leave the door opened by 3-4 inches, you can understand that it gets quite warm in there.

A64 3200+
2x Raptor 36GB
2x WD Caviar 40GB 7200s
Eventually a 9800 Pro or X800 XT
Powmax 500W PSU
2 optical drives.

Those are the noteworthy heat generators in the system. The problem isn't the CPU heat, the Koolance case does (reasonably, given the enclosed space) well enough keeping it under 38C. However the CASE temperatures are in excess of 45C.

So here's my incredible idea: I'd like to build a custom face for the space above the file cabinet on the (left side in the above link) and custom watercooling kit, drilling holes through the side of the desk to draw in the air, putting the entire PC inside the spot above the file cabinet with only the facia of the optical drives (and maybe a VFD display) showing through.

I was hoping perhaps the readers here could help me.

My goal is to have the major components all watercooled, overclocking is not a requirement (yet :blush: it always ends up being one though hehe), but most of all quiet.

The C-Systems pumps that I came across after reading your forums seem like an obvious choice for the pumps, but I have some questions.

The most dangerous question is: Should I watercool the Northbridge? The passive cooler on the MSI K8T Neo FIS2R seems to get incredibly hot.

I'd like to cool the CPU, with the capability of the waterblock I use to be transported to the 939 pin CPUs whenever I can afford to get one. In the past I've used the BIX and that seems to work well enough (gave my XP2100 in the Lian-Li case an overclock up to 2189MHz at 42C).

However I also want to cool the video card.

Should I purchase two BIX's and four C-System pumps for maximum flow rate?

Would two C-System pumps be plenty to acheive the idea flow rate in a "Multiple Block/Multiple Rad" setup in parallel configuration?

Frankly, Cathar's blocks seem incredibly impressive to me but I'm not sure what kind of flow rate the proposed setup I'm working on would give me and how that would hold up with his blocks. If I remember right, the Maze 4 does well at higher flow rates (for a commercial block - I realize Cathar's blocks outperform virtually everything at all flow rates)?

Would I be better off just buying the BIX2 (feel free to add-in other options for radiators, but bear in mind I have no way of working with metal reliably to make my own barbs) and running Pump-Rad-Cpu-Vid-pump? I thought perhaps running two independant loops would give each waterblock the coldest water available.

I'm sure that I can come up with more questions later, but I think this post has gone on long enough.

Sorry about the length.

-Jas

**EDIT**

If I did cool the NB, that would be part of the Video waterblock loop in the proposed two radiator/four pump solution. The fans I am likely to use would be the Panaflo FBA12G12L1A seen here.

[Trekari]

Perhaps isolating the questions will help:

1) Watercooling the Northbridge, yes or no? Seems like the last thread on that turned rather ugly and had lots of opinions.

2) Two separate BIX radiators (any other suggestions for a great radiator that doesn't require metalwork to affix 1/2" ID tubing?) or one big BIX2?

3) Two C-Systems pumps in parallel, one pumping through the CPU block and the other pumping through the video card/NB block, or four pumps - two for each loop with two radiators?

4) Unless the Cascade is available somewhere that I'm not aware of, is the TDX currently the best performer for high-flow rate setups?

I wouldn't ask #4, but I'm not entirely sure how to read the graphs on review sites. The deltaT between water and CPU is better when it is lower, meaning the water is absorbing more and more heat, right? :shrug:

Also...a fifth.

5) Would two C-System pumps in series pumping to just a CPU block such as the Cascade or TDX perform better than two working in parallel?

[bigben2k]

I'm sure the arguments against water cooling the NB were around the extra performance you'd get from it, which would probably be next to nil. However, having any electronic component run hot isn't necessary, and the general rule of cooling it by 10 deg C to double the life still applies (roughly). I'd cool the NB, no question about it.

You'd do well with any heatercore, where the only work involved includes trimming back the existing pipes (a simple hack saw job). BIX are pretty and fit nicer.

Pumps in series, one loop.

Yep, can't go wrong with a TDX. There are many, many acceptable choices to.

Drop pumps in parallel, you're wasting pressure.

[Trekari]

I appreciate the response greatly.

1) Pumps in series, one loop. - What kinds of cooling performance difference would I see in delivering the hot CPU water to the NB, and subsequently the GPU? Is it better, if I can spare the cash, to split the loops, or is the temperature difference between the two ideas not worth the price?

2) BIX single, or dual-fan? (BIX2)

Ideally I want the maximum cooling and quietest operation, lol. Two things I know don't hold hands very well.

Last question would be this then:

If I did go single-loop, would a 2gpm flow rate still be attainable with three blocks and a radiator? That is where the TDX performs best according to the review.

Remember this will be almost entirely enclosed inside a desk, with the only outside air being what comes in through the radiators.

-Jas

[bigben2k]

You want the highest flow, and to get there, you'll have to have enough pressure to cram that much coolant through all the obstructions. Pumping in parallel won't do that as nearly as well as pumps in series.

I'd rather advise you to chop up an existing core, without any soldering. It's easy to do, will give you more performance, and you can maximize the available space you have for a core. Ref mine (in sig). I really don't want to say anything else about BIX.

It's possible to balance performance and noise, it just comes down to what level of noise you can live with. In your case, since the unit will be enclosed, you have that much more to play with.

2 gpm is hard to do. Got a link to the pump?

If I was you, I'd consider having an air inlet as well as an outlet, going through that desk.

[kronchev]

two rads and 4 pumps is waaay overkill for your application. im oc'd 500 mhz, 0.30 volts, and i only have one small pump, and one rad. dT never is more than 12C and thats at prime95 loads. thats only with the processor however. i could probabily throw in my 9800Pro and see maybe only a degree or two increase (someone correct me if im mistaken). northbridge, always nice to have cooled, but never necessary.

[Trekari]

Originally Posted by bigben2k

You want the highest flow, and to get there, you'll have to have enough pressure to cram that much coolant through all the obstructions. Pumping in parallel won't do that as nearly as well as pumps in series.

I'd rather advise you to chop up an existing core, without any soldering. It's easy to do, will give you more performance, and you can maximize the available space you have for a core. Ref mine (in sig). I really don't want to say anything else about BIX.

It's possible to balance performance and noise, it just comes down to what level of noise you can live with. In your case, since the unit will be enclosed, you have that much more to play with.

2 gpm is hard to do. Got a link to the pump?

If I was you, I'd consider having an air inlet as well as an outlet, going through that desk.

http://www.procooling.com/reviews/ht...e_tests_-_.php

The pump I suppose is most attainable (since the hydrothruster looks rather huge) is the AquaXtreme 50Z-DC12. I was thinking about two C-Systems pumps in series, but I'm mostly concerned with the actual flow rate I can expect to get if I put a TDX Nozzle #4, Maze 4 NB block, and Maze 4 GPU block in series.

I'm a bit confused about the flow rate issue. Robotech's review here:

http://www.systemcooling.com/cs_pump-04.html

How is that data supposed to be interpreted in this scenario? His graph shows that two pumps in parallel give higher flow rates. I don't get it.

Perhaps the best way to go would be the TDX, GPU, then NB, pump, radiator since the two primary items I want to cool are the CPU and GPU - the NB is simply 'bonus'? I'm still (see above) unsure why it is better to put them in series and single loop versus a pump for the CPU, and a separate loop for the GPU and NB?

Last question then would be standard, single-fan heatercore, or dual-fan?

[pHaestus]

My goal is to have the major components all watercooled, overclocking is not a requirement (yet :blush: it always ends up being one though hehe), but most of all quiet.

If this is your goal then the Black Ice Extreme radiators (either one 120mm fan sized or 2) are not the best choices. I'd look at the 5x10 radiator sold by Becooling or a Black Ice 2 PRO (the HWLabs radiator sized for 2 120mms but thinner) along with a custom shroud and two undervolted quiet 120mm fans.

[Trekari]

Are you suggesting the Pro because it is less restrictive of airflow?

That seems to be the only difference between the BIX2 and BI2 Pro, unless I'm missing something else?

Thermochill HE120.2 seems to be along the lines of the BIX2 - 50mm thick, therefore also no-good?

**EDIT**

Ah, looks like the BI2Pro is also a 4-pass design versus 2 pass, longer time spent in the radiator means colder exit temps I would think.

[kronchev]

Originally Posted by pHaestus

If this is your goal then the Black Ice Extreme radiators (either one 120mm fan sized or 2) are not the best choices. I'd look at the 5x10 radiator sold by Becooling or a Black Ice 2 PRO (the HWLabs radiator sized for 2 120mms but thinner) along with a custom shroud and two undervolted quiet 120mm fans.

is that 5x10 any good? that looks pretty sweet.

[Trekari]

I'm no expert, but my concern with the 5x10 would be that it looks to be made of aluminum?

Isn't Al and Cu in close proximity a bad thing?

[Butcher]

I cool my NB - mainly because water is much quieter than air. Especially with the stupid little fans they use for NBs in general.

[pHaestus]

That 5x10 has copper pipes with turbulators inside. Very much underappreciated for quiet PCs.

The Black Ice Extremes are a lot more money for not much more cooling if you are using a quiet fan

"Ah, looks like the BI2Pro is also a 4-pass design versus 2 pass, longer time spent in the radiator means colder exit temps I would think." This is incorrect on several levels. HWLabs made a Black Ice 2 that was 4 pass but I don't think the Pro2 is. And the comment about longer time being related to colder temps is not correct. Would the ideal situation then be NO flow? Not quite.

[Trekari]

What I meant by that is a 4 pass design would spend more time inside the radiator than a two pass design at the same flow rate.

I know flow rate is important, I did not intend to imply that slower flow rates equal better performance. I may not know scientific data about watercooling, but I'm not /that/ naive.

-Jas

[kronchev]

Originally Posted by pHaestus

That 5x10 has copper pipes with turbulators inside. Very much underappreciated for quiet PCs.

i may look into that next time I set up a wc system...

[goat_guy]

What do you mean by turbulators?

Brian

[redleader]

Originally Posted by Trekari

What I meant by that is a 4 pass design would spend more time inside the radiator than a two pass design at the same flow rate.

I know flow rate is important, I did not intend to imply that slower flow rates equal better performance. I may not know scientific data about watercooling, but I'm not /that/ naive.

-Jas

But adding more passes does not cause the water to stay in the radiator longer. Think about it and you'll see why.

What do you mean by turbulators?

Something to mix up the water as it passes through. Improves performance somewhat.

What kinds of cooling performance difference would I see in delivering the hot CPU water to the NB, and subsequently the GPU?

Hot is a relative term here. You're looking at about one degree difference between the hottest and coolest parts of your loop with 1/2 tube and dual pumps most likely. Water holds a LOT of energy without getting much warmer.

[Trekari]

I guess that I don't understand what "4-pass" and "2-pass" are actually referring to.

I thought it to mean the water went back and forth across the face of the radiator four times, or two times. Naturally that would mean more distance to travel in a 4-pass design, which means longer time spent in the radiator.

What am I missing?

Also, a tangent here, but could someone perhaps explain what C/W really is and how to determine a rough estimate of coolant temperatures after each waterblock if you know the amount of heat the block is sitting on top of?

C/W Calories/Watt? i.e. if I have a 180Watt heat source sitting on a .17C/W waterblock, is 30.6 Calories being transfered into the water?

I'm likely wrong and initially I felt like I couldn't care less about the math aspect of watercooling, but I can't stand not understanding things. :$

[pauldenton]

Originally Posted by Trekari

I guess that I don't understand what "4-pass" and "2-pass" are actually referring to.

I thought it to mean the water went back and forth across the face of the radiator four times, or two times. Naturally that would mean more distance to travel in a 4-pass design, which means longer time spent in the radiator.

What am I missing?

Also, a tangent here, but could someone perhaps explain what C/W really is and how to determine a rough estimate of coolant temperatures after each waterblock if you know the amount of heat the block is sitting on top of?

C/W Calories/Watt? i.e. if I have a 180Watt heat source sitting on a .17C/W waterblock, is 30.6 Calories being transfered into the water?

I'm likely wrong and initially I felt like I couldn't care less about the math aspect of watercooling, but I can't stand not understanding things. :$

C/W = (degrees)C per Watt.... afaik

[Butcher]

4 passes tends to mean each pass is done more quickly. Assuming the radiators are of similar size then the water will take roughly the same amount of time to flow through them (flowrate is uniform throughout the system). That means the water in the 4 pass has to move approximately 4 times as quickly as in the 2 pass. 4-pass tend to be higher restriction which impedes flow rates which is why 2- and 1-pass are favoured.

C/W = degrees celcius / watt. If you have a 180W heat source and a 0.17C/W cooler then the heat source will experience a 30.6C temperature rise over ambient.