Koolance HXP-135 in a radbox - separating cooling loop from PC

[Vatharian]

Hi guys and girls!
As long time reader, I'm Xtremely happy to finally becoming part of this community :)

I'm asking for advices, possible traps and other info while planning my next LC upgrade.
First, I'll post my relevant specs:
- Core i7-2600K@ ~4.8-5.0 with Koolance CPU-360,
- Asus P8P67 Deluxe, no w/b
- 2x GTX570 with PCB v1 (single-slot capable ones), cooled by Aquacomputer AquagraFX blocks,
- third GTX570 is coming from RMA, and no block for it yet, has to contact AC if they still manufacture V1 nickel ones.

And yes, I know, that board is not 3-way SLI capable, 3rd will be dedicated to PhysX ;)

No O/C on GPUs, I'm kinda scared by nVidia cutting 1/8th of GPU from 580 and 1/3 of VRMs
- BeQuiet DarkPower Pro 1200W P7 (I *hope* it will be enough - it goes well with 2 GPUs),
- one SSD thrown behind mobo.
accessories:
- Koolance tube res,
- EK-DCP 4.0 (Jingway 1200),
- XSPC x2o 750 pump, lying around, broken res.
- 2x D5 (Alphacools VP655) with EK dual D5 single-loop top (due to design errors I had so restrictive loop, that this combo got a C-C-C-Combo breaker and I got 10 litres per hour total flow). Now disassembled.
- four 360 rads: two Phobyas V2 (mounted with sandwiched fans in front bays), one V1 (waiting in a closet) and EK Coolstream XT (lying on top of case), one 120 rad (also lying aside).

With this kind of setup I have very decent temps, like 33-40 water and up to 60 on both GPUs and CPU, but it's both noisy (10 fans total) and cramped and I lost all my front 5.25" bays:

127165

127166

So I decided to go nuts and build radbox. Then it evolved to full, independent external cooling unit.
Overview, not completed setup - outlets are missing, and of course alot of drainage, misc plumbing and electrical stuff is not yet ready:

127167

127168

When I finish the models I'll make sure to update these drawings.

127259

Inside are two completely separate loops, one is strictly closed inside the radbox and goes:
res -> pump -> heat exchanger -> 3 rads -> res
second one is 'open':
inlet -> res -> pump -> heat exchanger -> outlet
On the PC side I'd need only to connect water blocks to the outlet and voila, it's ready.
Estimated total cost: 1000?‚? or $1100, around half of that with already owned hardware.

Pros: Filling/draining loop with a flip of a switch, and temp/flow monitoring with ease. Very low flow resistance.
Cons: There will be some trouble with filling, since it will be LOWER than PC. And size, front panel should be roughly size of a standard home-theater amplituner, but twice the depth. I have space for it, so no big deal.

And here comes, what lies in the basics of this idea: The Heat Exchanger. Namely, the Koolance HXP-135.

http://koolance.com/index.php?route=product/product&product_id=943

127169

To be honest it's almost identical to what is being commonly used in home heating, save the barbs which are 1/4", not 3/8" or 1/2". And, of course, it has some bling on it. It can transfer insane amount of energy between two liquids, up to 30 kW with ease. Koolance advertises tenth of that and I fear there is a reason to that. Any heat exchanger works best, when there is good difference or delta between temps. In PC cooling we talk about fairly small differences. I spent days trying to get any data on efficiency in low-deltaT environments, and completely and utterly failed.

I'm asking you for advice - is this going to end in a disaster (i.e. 15*C difference and low heat transfer), or did I just came up with something ingenious?
Do anyone has experience in this kind of cooling (or heating!) with plate heat exchangers?
What are best practices while building/designing external radboxes?

Any suggestions and explanations will be welcome!

Thank you all
V.

PS. I've used Sketchup components from here:
http://online.freeware.info.pl/games.html

EDIT: Added schematics for the radbox setup.

[LeoAndrade]

Nice project you have there! Although, I have no numbers about the performance of this heat exchanger. It will be the key of your project.

I think a single, all together loop would be safer. People work with radboxes for long time, and everybody knows that it can offer a huge cooling performance. The introdution of a heat exchanger WILL reduce the cooling efficiency, but how much I don't know.


Wish you luck!

[Vatharian]

I intend to build such ECU (cool! External Cooling Unit!), mainly because of simplicity of using outer loop. Bleeding 3 or 4 rads will always be pita. When they're prefilled - it's just fun!
The temps doesn't necessarily need to be higher. PHEs do represent some barrier in cooling, true, but there are other factors:
- I have two rads sandwiched, and they are in wrong order, so only the top one is getting enough airflow. Switching to all-flat radbox will give some efficiency boost
- I have restrictive loop, not only CPU-360 is a flow killer of prima sorta, as you probably know for yourself, but all sorts of sharp 90 degree turns I had to make decrease flow. Switching to very, very high flow on radiators will further bring down the temps, while doing so with waterblocks will have minimal impact. delta T increases - more cooling goodness.
I only wish there was a way to do the math beforehand.
http://uslugi.nie-spamuj.eu/rtv,1,0.html
What do you mean by 'safer'?

[LeoAndrade]

Everything you said is corret, and again, doing the math before the project would be good.

What I tried to say with "safer" is that working without a heat exchanger is usual, so the cooling performance can be predicted.

[mele]

Hummm.......this gives me an idea.......

[Vatharian]

Id love to count everything upfront, but there is no data. thats exactly why Im here for!

If anyone actually used the hxp, please respond!

*Drops a penny to a piggy bank*
I will ultimately build the radbox, with a direct loop, then drop the exchanger to check.

[bds71]

am i missing something? a heat exchanger is simply a radiator. even the one from koolance is just a radiator - stainless steel at that which tells me the cooling properties will be quite a bit worse than a normal/regular copper rad. what am i missing? is there some "active" component to this guy that suggests it will perform better than a copper rad?

note: sorry if i'm missing the obvious - i did look, but didn't see anything "different" other than its design...

[Vatharian]

Well, *almost* correct ;). Every radiator is a heat exchanger, as it transfers heat (energy) from one medium to another. But this one is designed to transfer heat from one liquid (say, loop 1) directly to other liquid (i.e. loop 2). Liquids in both loops do not mix. Greater the temperature difference and greater the flow, greater heat transfer occurs. The best use of such device would be having normal liquid loop, and by incorporating PHE into the loop you can route cold tap water, or any artificially chilled liquid, i.e. freon.
I'm intending to use it to physically separate part of loop containing water blocks from part with radiators. It's for convenience, nothing more. I could actually ditch whole thing and use pair of quick-disconnects, but you know, if you're into liquid coolling, there are seldom clear answers ;) http://mailing.nie-spamuj.eu/e-mail.html
Of course there are other uses: I could think of load-balancing two loops - say you have 2 or 3 GPUs and powerful CPU and two separate loops. By routing both of them through HXP-135 you could use some of cooling capacity of the other loop in case only one is stressed. And still have two colors in!

[Sc.D. Puschkin]

http://imageshack.us/photo/my-images/560/img0127fn.jpg/
http://imageshack.us/photo/my-images/845/img0124uz.jpg/

Have absolutely no problems ;)

[bds71]

OK, so instead of transferring the heat from the water to the air (as is the case with a normal rad) this unit transferes the heat to another liquid - requiring, at a minimum, another water loop. using this example (rather than exotic sub-zero/compressor type loop): the second loop (the cooling loop if you will) still has to exhaust the heat "somewhere." if its a regular copper rad water loop, it exhausts to the air. in this instance, you just transferred the heat from one water loop to another via this heat exchanger - but still got rid of the heat in the same manner. is there any advantage to this? i still don't see what purpose the heat exchanger is serving that the second loop couldn't do by itself? even if you have an exotic type of sub-zero loop to cool the water, couldn't you simply use that loop to cool your system rather than a heat exchanger to transfer the heat from one loop to another? eventually, you'll still hit your (heat) saturation point - the question is: will this have any effect on OCing? stability? dT? room temperature?

again, not knocking what your trying to do. just trying to figure out why your doing it this way, and what benefit it will have :) lord only knows i've toyed with the unconventional - i'm a TEC user - need i say more lol :)

[Vetalar]

on product page of this PHE in tab "files" there is plenty of performance graphs for different flows and deltas. take calculator and get performance numbers You want - no rocket science IMHO

[Vatharian]

You've just very accurately described what I'm intending to do.
By having these two loops separated, I'm accomplishing a bunch of things:
- I can fully contain the "cooling" loop with radiators inside the radbox, making it essentially maintenance-free,
- I will have 'external' loop, that has everything needed to run it with inlet/outlet on the chassis of radbox. I can short-circuit it and it will run (I know it's idiotic), and inside the computer I need only to connect waterblocks to in/out ports and i'm done. Filling and bleeding external loop will be done using plumbing and controls inside the radbox.
- I'm removing all the hassle of bleeding radiators. With enough flow you can essentially blow the air out of waterblocks, but in most cases it can't be done with rads. You WILL have to perform shaman's dance with it. With 80+ lbs case it's both risky for equipment and definitely leading to spine injury. Rads are filled completely, in one shot, forever.
- I can go wild on what coolant to use inside the case, where it will be seen, and leave distilled water with Biocide and anti-corrosive in he radbox where it will NOT be seen. Just a drop of dye for leakhunt.
- I'll always know what kind of flow I'm getting in cooling loop. Rads are mildly restrictive, but they DO like higher flow. Otoh external loop will not be restricted by rads and I can trim the flow for best results (go over 2500 l/h and introduce cavitation to most of CPU blocks, shooting down ANY heat transfer).

heat exchanger WILL be barrier for cooling efficiency, but weighting all the factors, I see a purpose and still think that actual cooling capacity will increase.

running TEC is like fighting fire with fire, imo. Still pretty cool tho ;D

[bds71]

OK, i see it now - not necessarily a performance move, but one of convenience/making it easier to maintain. as for cooling capacity: with two loops i would say you get the total "capacity" of both loops combined (which essentially equates to a large reservoire) meaning the loop will take longer to stabilize, but the ending dT prob won't change much) also, with the stainless steel heat exchanger between the loops, you may actually lose some performance (?) but, maybe not...something to look for at any rate

as for TEC's - highly underrated...they're making a come-back, just you wait and see <j/k>....nahhh, i love em - so much potential for some really cool stuff - if they (the manufacturers) could just "get it right"

well, anyway, good luck on your project, and keep us posted :)

[Buckeye]

With those heat exchangers I think the temp dif will be to small to really have an impact on anything.
Like others have said they are used to transfer heat from one loop to another, they did not mix.

If for example you had a loop that was built around a chiller or a seperate loop going to your pool then you would see a better delta, for sure with a chiller.

Cascade builders use those all the time where they are transfering heat from second stage to first stage because they are using different gasses so they cannot mix. But there are other controls used like TVX's and things like that to control gass flow.

I have seen that heat exchanger on Koolance's site and wondered what the heck do they use that for in water cooling. Maybe in a server room cooling setup where the server is one loop and the other loop is feed into a chiller, I don't know.

Inside plates are not nickel plated I don't think so even Stainless Steel will rust and mix metals in a loop, could be bad news.

Channel Plate Material: SUS304 Stainless Steel
•Brazing Material: Copper

[Vatharian]

Buckeye:
I have couple of Koolance plugs covered with rust, so that's really bad news. So far I only watched for aluminium in the loop. I'll ask my dad, he should know how this specific steel reacts with water (he's marine enigneer).
http://nie-spamuj.eu/email.php
bds71: The 'external' loop has zero cooling capacity - it gets all the heat transferred by PHE. I'll preprare some quick schematics and add it to first post. Should've done it in the first place...

[LeoAndrade]

Copper brazing should be enough to prevent any galvanic corrosion scenario. You can do a research about that SUS304 steel to see it's chemical and mechanical proprieties though.

[Vatharian]

copper brazing... Well, yet again I find my English not good enough.
http://translator.telewizor.eu

@mele: I completely missed your post!
What do you have in mind?