My new WC loop - pics, new ideas, & $$-saving tips

Dew itt right

2[H]4U
Joined
Oct 28, 2005
Messages
3,314
EDIT: The WC setup below is now listed on the FS threads HERE...

Well, this is my third water loop build and I definitely have to say it's the best and most thought-out configuration so far. I did a few things and used a product or two that most may have never seen used in this application before that turned out to be perfectly suited for watercooling. I'll also throw in a few ways I saved quite a bit of money so if you're getting ready to build your own loop – listen up!

Disclaimer said:
First things first: The 8600GT SLI setup -- Yes that's right, I have (2) 8600GT's in SLI. I get crap about this all the time. “Why'd you waste your money?! You should have gotten a 8800! They're the best video cards in the world! They make two 8600's cry like little girls! Blah blah blah...” Actually I had an Evga 8800GTS 640Mb Superclocked. I sold it and picked up these two cards for more than $100 less than the single 8800. The main reason for the switch was because, as a general rule, I try to never have any component over $200 in my case. This way if any given part breaks, I know I won't be out more than $200. This is especially true when watercooling because crap happens! Plus if one GPU breaks down I still have another to keep the system up and running. I'm very pleased with the performance of these two cards as well. I know benchmarks aren't the end-all say-all in performance, but these two cards OC'd on air did 9700 in 3DMark06 whereas my old 8800GTS Superclocked only did 9400 or so at stock (superclocked) settings. Plus I now have 1Gb of GPU memory instead of 640Mb. So please don't give me any crap about my SLI setup – I've heard it all before and I don't regret the switch one bit.

The External Setup...

WC-External.JPG


This idea started simply because I didn't have any room inside the case obviously for a 3x120mm radiator. Once I pick a spot out for it on the right side panel I thought to myself, “self, why don't you just put all the guts back here to free up space in the case?!”. After patting myself on the back for such a great idea, I began figuring out how the parts need to be laid out...

1) The radiator -- The radiator was mounted horizontally on the bottom of the panel simply because it was the only convenient spot to pass through to the interior of the case. Plus, having the water flow up thru the radiator instead of down help push the air bubbles on through to the reservoir instead of forming air pockets in the radiator.

2) The reservoir -- The reservoir was mounted with the barbs facing the left to prevent the tubing from ever having to cross itself to get to the pump. I'm not sure how most people configure their inlet/outlet ports on this res, but I made the middle port the inlet to help stop air from collecting under the divider (and obviously to keep from sucking air out to the pump) and I left the top port capped for use in filling and air bleeding.

3) The pump -- The pump was mounted below the reservoir to allow gravity to help feed the pump from the reservoir. This setup was originally done for a MCP355 to be mounted to the side panel but I decided to spend the extra money on the MCP655 to have flow control and a more quiet rig. This pump is secured to the top of the radiator with velcro (which is an extremely good vibration dampening material due to the loose connection).

4) The grommets -- As usual, grommets were used to allow the tubing to pass through the case. However with the pump and radiator fan outside, I also had wiring that needed to pass through. For this, I bought the smallest grommets I could find and used these small grommets to pass the wiring through the case. As you may see, I did have one extra grommet at the radiator outlet. My original intention was to pass back into the case. They do, however, have plugs for these types of mistaken holes at hardware stores, just haven't gotten around to it yet.

The Manifold...

WPCM2-M3-08B.jpg


I've seen pics of people using this concept of splitting the supply line with a series of tee's but I've never seen somebody actually use a real one-piece manifold with valves on each line. This creates innumerable possibilities as far as expansion and customization are concerned. The benefits...

1) Uniform fluid supply temps -- Separating all three supply lines means each water block is receiving the lowest temperature fluid it can receive as opposed to piping them in series where the heat from one block warms up the water and then runs through another block and another. This way the fluid is cooled by the radiator and goes directly to each individual block. Yes splitting the lines means lower flow to each individual component, but from my past experiences I noticed that supplying the waterblocks with cooler fluid does much more to lower temps than giving the block more fluid flow. Besides, this is kind of where my next point comes in...

2) Ability to balance each branch -- With each supply line having a valve, this allows you to partially close each valve in order to direct more flow to one area and less flow to another. For example, you not be very big on CPU OC'ing but you're a huge gamer! So you'll want to leave the GPU supply line fully open and partially close the CPU and chipset lines to direct more flow towards the video cards. (NOTE: The temperatures in my “results” section below are with the system unbalanced – all vales are fully open. You'll notice the CPU needs more fluid and the GPU's have fluid to spare.)

3) Temporarily concentrate fluid on one particular component -- With my 680i board, I can monitor the CPU, chipset (in BIOS), and GPU temps and balance the water until the temps are where I want them. Or if you want to find the max OC on your CPU, you can temporarily divert more water to the CPU to cool it better. Then when you want to max out your FSB or GPU you can again temporarily divert water to where it's needed most.

4) Less likely to leak -- You could make this manifold from 2 tee's, 1 elbow, and 3 valves but you'd be looking at making 14 connections where with a one-piece manifold you only make 4 connections (or 6 if you count reducing the inlet connection)

5) Room for expansion or a drain line -- If you're only using three branches now but want to use a fourth in the future, just buy a 4-way manifold at the start of the project and leave the valve closed on the last unused branch. Personally, i wish I would have bought a 4-way manifold and setup the unused port on the bottom as a drain line to eliminate the need to break a connection to drain the fluid.


The manifold (LINKED HERE) has (1) 3/4” PEX inlet and (3) 1/2” PEX outlets. The 3/4” PEX connection fits 5/8” ID tubing perfectly and the 1/2” PEX connection fits 3/8” ID tubing. To keep my loop all 3/8” ID I purchased a brass 3/4” PEX x 1/2” PEX adapter (at Home depot) and used a small (2” or so) piece of 5/8” ID tubing to connect the adapter to the manifold. This allowed all four connections to be with 3/8” ID tubing.

The whole shebang...

Well, now that the system is complete, here's a list of all the components it took to make the loop...

-(1) Swiftech MCR320 radiator (LINK)
-(1) Swiftech MCRES reservoir (LINK)
-(1) Swiftech MCP655 pump (LINK)
-(1) Swiftech Apogee GT Extreme CPU waterblock (LINK)
-(1) Swiftech MCW30 chipset waterblock (LINK)
-(2) Swiftech MCW60 GPU waterblocks (LINK)
-(2) Thermaltake CL-C0025 copper vga memory heatsinks (8-pack) (LINK)
-(1) Watts WPCM2-M3-08B copper PEX manifold (LINK)
-(1) 1/2" PEX x 3/4" PEX Brass Coupling (LINK)
-(1) 32oz. bottle of PrimoChill Ice Invisible Blue UV Reactive Non-conductive coolant (LINK)
-(10) Feet of Excelon RNT 3/8” ID 5/8” OD tubing (LINK)
-(3) 5/8” ID rubber grommets (LINK)
-(2) 1/4” ID rubber grommets (LINK)
-(10) polypropylene 3/8” elbows (LINK)
-(4) polypropylene 3/8” tees (LINK)
-(2) polypropylene 1/2”x3/8” reducers (LINK
The Results...

Testing conditions:
FSB: 415MHz
CPU: 3320MHz
RAM: 1,066MHz
GPU cores: 620MHz
GPU mems: 850MHz
--2 instances of SP2004 running simultaneously (Blend and Small FFT's)
--ATITool running artifact check

AIR COOLING (all fans at full speed)
...................idle......................load....
CPU............42.........................67.....
GPU1..........53.........................73.....
GPU2..........55.........................73.....


WATERCOOLING (radiator fans full speed)
...................idle......................load....
CPU............35.........................58.....
GPU1..........38.........................46.....
GPU2..........39.........................45.....

WATERCOOLING (radiator fans off)
...................idle
CPU............48
GPU1..........49
GPU2..........48

**The tests and temps above were done with all of the manifold branch valves fully open. As you can see, I need to divert a little more fluid towards the CPU. The GPU's are being cooled extremely well – a 27 degree drop from aircooling. I'm hoping to balance the temps out with the valves. It's too bad the I can't monitor the chipset temps. If it's monitored in the BIOS I may be able to just balance it based on idle temps.

Tips & Other Notes...

1) Tubing
-- Instead of spending a small fortune for tubing at a PC website, try ebay, amazon, and laboratory supplies websites. I picked up a 100ft roll of Excelon RNT tubing (as good as if not better than Tygon) for $30 shipped! Then I turned around and sold what I didn't need for $0.75 which is still an awesome price for high quality tubing. I wound up actually making money and keeping 20ft for myself. Also, USPlastic.com sells MANY different types of Tygon tubing including the Tygon R-3603 that FrozenCPU sells except where it's $3.00/ft at FrozenCPU and most other PC-mod retailers it's only $1.49/ft at USPlastic.

2) Need a case window? -- After installing a sweet looking WC loop, you'll want a case with a side panel window to show it all off and to keep an eye out for leaks. When I bought my Cooler Master Centurion 5 a while back, I bought the model without the window because I wanted to install fans on the side panel. If you're ever in the same boat I was, before buying a do-it-yourself side panel window kit (and the tools needed to install it), check to see if the manufacturer sells and windowed side panel that you can buy by itself. I wound up finding my windowed side panel HERE at SVC.com for $9.99! It was cheaper that those DIY kits, obviously required less work, and looked more professional in the end.

3) Don't spend a fortune on plastic fittings -- At the PC modding websites you'll pay $1.50 or more easily for a single plastic elbow or tee fitting. I bought a couple dozen fittings at USPlastic.com for about 25% of the price! Most of the fittings I bought were around $0.40 a piece and you can choose between different types of plastics and different colors. They even have threaded fittings that I bought for my swiftech radiator although these are very prone to leaking because they do not have the lip above the threads to hold an o-ring down. Also know that the natural polypropylene fittings are UV reactive (a bonus I didn't know about until I installed them).

3) Going UV? Get rid of the LED's! -- The UV light cast by UV cold cathodes is typically very soft. So soft that even a few LED's can wash out the glow from your UV reactive components. Regrettably, the blue leds shining on the motherboard and the red and green flashing LEDs on the top of my Crucial Ballistix Tracers really drown out the UV effect of my lights. I may just jump on a pair of regular Ballistix if I find a good deal. If you want a bold glow from your UV reactive components, get as many of the LED's out of your case as possible.

4) Making the MCW60 and 8600GT compatible -- At the time of this writing the only way to watercool an 8600GT is with a full coverage EK block that costs almost as much as the card does! The MCW60 is not listed as compatible with the 8600's, however the stock nVidia bracket's mounting holes line up perfectly with the 8600GT. The only problem was that the legs on the mounting bracket are a pinch too long and when mounted leave the surface of the block slightly off the GPU chip surface. The only ways to fix this are to (A) shorten the legs by grinding and hoping you don't screw up the threads inside too much or (B) to separate the mounting bracket from the block just enough to shorten the standoff distance from the bottom of the legs to the waterblock surface. I chose (B) so as to not permanently alter the waterblock.
. First unscrew the four screws on the top of the block, separating the copper base, derlin top, and metal bracket. Take one of those screws with you to your local Home Depot/Lowe's and find the fiber washers in the hardware isle in one of those tiny drawers. Get the smallest fiber washer their that the screw can pass through. Then reassemble the water block putting the washers between the derlin top and the mounting bracket. The washers won't sit flat, but when you start screwing the screws back in the washer will form to the contour of the derlin top and you'll have a nice tight fit. This should raise the mounting bracket off the block just enough to help the copper case make contact with the GPU chip surface. And don't forget to buy some memory heatsinks for the memory chips.

Finally, Some Pics...


WC-Open_side.JPG

.
.
.
WC-Open_side_enlarged.JPG

.
.
.
WC-CPU_NB.JPG

.
.
.
WC-SLI.JPG

.
.
.
WC-Side_window.JPG

.
.
.
WC-Whole_case.JPG
 
Last edited:
Nice build :)

About the tubing tip, it's great and I was one of your buyers. I'm pleased with my 15 feet and I might buy more from either ya or eBay (too bad it will cost me 40$ shipped to my door due to high canadian shipping price).
 
I must say that is a very original design!

Only thing I would have done different is used "Y" fittings for combining the loops on the return side and maybe had the CPU on it's own return. Other than that, GREAT WORK!
 
I must say that is a very original design!

Only thing I would have done different is used "Y" fittings for combining the loops on the return side and maybe had the CPU on it's own return. Other than that, GREAT WORK!

Yeah, I realized that when I was putting it together but it was obviously too late at that point. I was also surprised by how close the valve outlet was to the supply tee above the first GPU. It's a very tight tubing bend (there wasn't even enough room for an elbow) but the tubing was so sturdy that it doesn't kink at all. Definitly couldn't have done that bend with most tubing...
 
Nice job. I have to point out though that your superstition of a coolant temperature advantage in parallel is just that, superstition. If you were to actually measure the difference in temperature in a serial loop you would be hardpressed to find more than a 1-2C difference in just about any loop. The reduced flow will almost always have more of an effect than slightly reduced temperature. Then again, you probably won't notice one way or the other unless you have a very extreme setup.

When done correctly parallel configurations can provide perfectly good, sometimes better performance in certain situations. Another advantage that parallel sometimes provides is cleaner tube routing.
 
Nice job. I have to point out though that your superstition of a coolant temperature advantage in parallel is just that, superstition. If you were to actually measure the difference in temperature in a serial loop you would be hardpressed to find more than a 1-2C difference in just about any loop. The reduced flow will almost always have more of an effect than slightly reduced temperature. Then again, you probably won't notice one way or the other unless you have a very extreme setup.

When done correctly parallel configurations can provide perfectly good, sometimes better performance in certain situations. Another advantage that parallel sometimes provides is cleaner tube routing.

I agree with that in the fact that these are such small scale applications that little difference can be seen either way. Take fluid flow for example, I can run the 1-5 scale speed setting on my MCP655 from 5 all the way down to 1 and see no significant temperature change (rendering an adjustable manifold fairly useless). We speak in hypotheticals when we start talking about "the best loop configurations" although I must, in my ignorance, hold to my belief that in my particular scenario piping my four components in parallel will provide slightly (perhaps unnoticeably) better cooling for the components as a whole.

Thanks for all the props as well...
 
Back to what Erasmus354 was saying...

The TDP (Thermal Design Power) of all four components (E6750, 680i, & (2) 6800GT's) combined is 231W and the max flow from the MCP655 is 317 GPH. Using basic heat transfer calculations, 231W applied to 317 GPH of water is only enough heat to raise the water stream's temperature by 0.18C! And that's assuming 100% power usage, maximum flow, and perfect heat transfer. I think you can conclude from this that it really doesn't matter how you arrange it, watercooling just works.
 
I have to say it looks good, but i am still concerned about the functionality, all i see in the tubing are 90* corners, and all i can think about is how performance is hindered due to it.
 
I have to say it looks good, but i am still concerned about the functionality, all i see in the tubing are 90* corners, and all i can think about is how performance is hindered due to it.

True (normally), but a pump pushing 5 gpm is WAY more than enough flow through a 3/8" tube! Trust me, when the pump is cranked up - that fluid is movin'! I typically leave it around 3 for acoustical reasons. Besides, you don't notice any real temperature difference between the pump's max flow and min flow (which are drastically different speeds). It's much more than enough flow for such a small application.
 
I have to say it looks good, but i am still concerned about the functionality, all i see in the tubing are 90* corners, and all i can think about is how performance is hindered due to it.

Elbows don't reduce loop performance to any appreciable degree. :eek:
 
Elbows don't reduce loop performance to any appreciable degree. :eek:

A bajillion and a half will hehe. Also elbows in certain places, directly before the inlet, have more of an effect than other places. However, if it makes for a clean setup I say go for it.

@Dew : Thats exactly my point, and what people don't realize. Water can absorb A LOT of energy before it increases its temperature by even 1C. The water in a typical loop is flowing so quickly that there is very little temperature difference.
 
I like the ideas and functionality presented in this loop - there is certainly something to be said for the ability to direct flow where you want it. Overall it's a very nice setup.

That said, it's been pointed out that those components could be arranged in just about any loop configuration and still provide the cooling that you need - case and point is the fact that you're using 26 connections with 4 T's and at least 3 elbows on the inside portion of your loop, and two elbows directly in front of your pump inlet, and you're still getting decent performance. If it were me, I would've probably just used a serial loop with 9 connections, zero T's and maybe 2 (copper) elbows on the inside portion of the loop if needed.

But then, my loop would be a lot less interesting. Nice work. :)
 
great design with the manifold. It's good to see somebody travel outside the conventional loop. I would love to see some flow tests with the manifold.. The MCP650 really seems to shine here with all that added restriction..
 
The MCP650 really seems to shine here with all that added restriction..

Quite the contrary. The benefit of going in a parallel configuration is that it LESSENS the restriction, not increase it. However, it usually doesn't lessen the restriction enough to overcome the fact that the flow is split between the different branches. So you will have a higher overall system flowrate, but each individual component in parallel will have a lower flowrate as compared to a similarly configured serial setup.
 
Quite the contrary. The benefit of going in a parallel configuration is that it LESSENS the restriction, not increase it. However, it usually doesn't lessen the restriction enough to overcome the fact that the flow is split between the different branches. So you will have a higher overall system flowrate, but each individual component in parallel will have a lower flowrate as compared to a similarly configured serial setup.

yes your right but I wasn't referring to the manifold or parallel design degrading flow, simply the high amount of T fittings/blocks used..
 
With a decent pump, deciding which flow structure (parallel or series) you want to cool your components is like deciding whether to kill a fly with a frag grenade or a nuke...
 
It looks really nice! I'd have ran everything in series and without all of the elbows and T fittings, but that's me. I try to minimize tubing length and restriction for the best flow rate, especially since I run a Storm CPU block which adds a lot of restriction by itself.

With a decent pump, deciding which flow structure (parallel or series) you want to cool your components is like deciding whether to kill a fly with a frag grenade or a nuke...

That is not necessarily true. If you use a Storm (or probably a Fuzion block with the nozzle upgrade), you want the CPU block right after the pump for make pressure to make it work more efficiently. Otherwise, yeah it shouldn't matter a whole lot.
 
... If you use a Storm (or probably a Fuzion block with the nozzle upgrade), you want the CPU block right after the pump for make pressure to make it work more efficiently. Otherwise, yeah it shouldn't matter a whole lot.

As far as flow is concerned it wouldn't matter if it were first or last in the loop. The system flow rate (in series) is constant through the entire loop and is based on the pressure drop over the entire system. Yes the pressure drop over the storm will be larger than most, but it will still get just as much flow as all the other blocks. Where you place the block in relation to the pump will change the pressure at the block's inlet and outlet, but regardless of what the pressure is at the inlet of the block, the pressure drop thru the block and the flow rate thru the block will remain constant.
 
Yeah, I know the elbows and tee's doen't exactly help my flow out, but I got the MCP655 so I'd have some flow to spare. I don't mind sacrificing a little flow for cleaner tube routing. It's looks alot better this way (in my opinion).
 
Elbows do add restriction, each one is equal to 4ft of tubing.
Two links you should check out:

The Elbow Showdown, Copper is King

One 1/2" Elbow = 4' of 1/2" ID Tubing (Test Results)

Yeah, and 4 feet of tubing decrease flow-rate in a way that leads to a 0.1'c temperature difference according to Marci. Now, if you have the ability to measure to 0.1 degree Celsius, more power to you, but while my thermistors are +/- 2c accurate, I'll stick with my elbows ;)

I said to an "Appreciable degree." After you add in a single elbow, the loss in flow thereafter decreases drastically. And, flow-rate isn't the be and end all of water cooling performance either, most people can turn their pumps down to the lowest setting without noticing an increase in temperatures. 3/8" tubing was proven only marginally worse than 1/2" tubing. So, I too value clean routing over flow-rate.
 
Yeah, I know the elbows and tee's doen't exactly help my flow out, but I got the MCP655 so I'd have some flow to spare. I don't mind sacrificing a little flow for cleaner tube routing. It's looks alot better this way (in my opinion).

That's where my thinking regarding what constitutes 'cleaner tube routing' completely departs from yours. However, like I said, that's a nice loop.
 
As far as flow is concerned it wouldn't matter if it were first or last in the loop. The system flow rate (in series) is constant through the entire loop and is based on the pressure drop over the entire system. Yes the pressure drop over the storm will be larger than most, but it will still get just as much flow as all the other blocks. Where you place the block in relation to the pump will change the pressure at the block's inlet and outlet, but regardless of what the pressure is at the inlet of the block, the pressure drop thru the block and the flow rate thru the block will remain constant.

That's 100% correct. Order of components doesn't matter at all.
 
As far as flow is concerned it wouldn't matter if it were first or last in the loop. The system flow rate (in series) is constant through the entire loop and is based on the pressure drop over the entire system. Yes the pressure drop over the storm will be larger than most, but it will still get just as much flow as all the other blocks. Where you place the block in relation to the pump will change the pressure at the block's inlet and outlet, but regardless of what the pressure is at the inlet of the block, the pressure drop thru the block and the flow rate thru the block will remain constant.

I misread what you were saying. :D Yes flow is consistent. Pressure is not. I was thinking of pressure, not flow. Hence my comment about pressure in my original post.
 
Yeah, I know the elbows and tee's doen't exactly help my flow out, but I got the MCP655 so I'd have some flow to spare. I don't mind sacrificing a little flow for cleaner tube routing. It's looks alot better this way (in my opinion).

The MCP655 is a great pump, and you're probably not going to see such a pressure drop that your tubing would cave (seen this happen with weaker pumps in less restrictive loops than yours), but it only has 10ft of head and after that the pressure drop is somewhat drastic and would affect your temperatures.


Yeah, and 4 feet of tubing decrease flow-rate in a way that leads to a 0.1'c temperature difference according to Marci. Now, if you have the ability to measure to 0.1 degree Celsius, more power to you, but while my thermistors are +/- 2c accurate, I'll stick with my elbows ;)

I said to an "Appreciable degree." After you add in a single elbow, the loss in flow thereafter decreases drastically. And, flow-rate isn't the be and end all of water cooling performance either, most people can turn their pumps down to the lowest setting without noticing an increase in temperatures. 3/8" tubing was proven only marginally worse than 1/2" tubing. So, I too value clean routing over flow-rate.

I have to disagree about the pump speed vs temperatures. I just recently switched to a Laing DDC from a D5 variable that was in a CPU-Only loop with a BIX240. Dialing my pump down from 5 to 3 would result in a few C temperature increase, and that's on a loop with very minimal restriction (one rad, one block, and maybe 3 ft of tubing at most).
 
I have to disagree about the pump speed vs temperatures. I just recently switched to a Laing DDC from a D5 variable that was in a CPU-Only loop with a BIX240. Dialing my pump down from 5 to 3 would result in a few C temperature increase, and that's on a loop with very minimal restriction (one rad, one block, and maybe 3 ft of tubing at most).
In an unrestrictive loop such as yours, the flow rate is more easily affected by a single component (there being only 3), so it isn't surprising that your flow could decrease a lot (for instance, nearly 0.5gpm) simply by dialing down the pump.

However, in a loop with more components and more restriction, the change in flow rate as you dial the pump down is going to be much smaller, mainly because the initial flow rate on setting 5 will be much lower, and as flow rate decreases, the pressure drop across each component decreases as well, resulting in a MUCH smaller change, and thus much smaller change in temperatures. Arcygenical's assertion is correct with regard to a restrictive loop.
 
Wow... seeing this CLEAN layout gave me some inspiration to get back into WCing... potentially for my next quad-core build... hrmmm I MIGHT just get back into the (addiction) hobby— been a while...:(

Let me ask you:

1) Ever thought about mounting the radiator on top of the case and let the venting case-air cool it / make a shroud to conceal it?

2) Did you ever measure your flow rate with that completed loop?

3) Ever thought about (for flare/looks) using brazed/soldered copper tubing for the external components? Give the tubing a good polish + lacquer coat and bling that mofo out? Yep! :D

Lastly: I'd be very interested to see the temps of this minus some elbows... I think it would be negligible and I say that for 1-5C... 5C sounds like a lot, but in reality if your whole OC is affected by 1-5C it's not stable in my op... 1-5C means nothing, except to numbers sticklers. But it would be a nice test to conclude...

Great job! Keep'em CLEAN! :cool:
 
Thanks! I didn't put the rad on top because I wanted it all out of sight behind the case. I never did get a flow reading because I don't have anything to take that measurement with but I know the flow is more than adequate. I keep the speed setting (1-5) on 2 and my E8400 peaks at 50C running full load at 4.0GHz. Cranking the speed setting lowers temps by 1-2C and doesn't really allow for any more OC'ing. If you want an updated look at it I just listed the whole thing for sale yesterday HERE...
 
Obviously a lot of work went into this build. But I think you made some questionable layout decisions. You mention running the pump on a low setting for acoustic reasons. This comment does not surprise. Placing a whiny pump like the D5 externally almost forces running the pump on low if wanting to maintain sanity. If you relocate your pump inside the case I think this would improve your build.

Secondly is the number of elbows, ouch for flow performance.

IMO parallel is fine for gpus and bridge chips, but better to split the loop downstream of the cpu.

On the positive side the res placement seems very convenient and nice job trying to make things look clean.
 
Obviously a lot of work went into this build. But I think you made some questionable layout decisions. You mention running the pump on a low setting for acoustic reasons. This comment does not surprise. Placing a whiny pump like the D5 externally almost forces running the pump on low if wanting to maintain sanity. If you relocate your pump inside the case I think this would improve your build.

Secondly is the number of elbows, ouch for flow performance.

IMO parallel is fine for gpus and bridge chips, but better to split the loop downstream of the cpu.

On the positive side the res placement seems very convenient and nice job trying to make things look clean.

Again, I have to disagree about the elbows. The amount of static pressure they add to the loop is negligible compared the to the pressure head the pump can handle. When the pump is turned up to 5 you can see and hear a definite sizable increase in flow and the temperatures don't drop much at (if any) so the flow rate is more than fine. Maybe it could be a problem for a weaker pump, but definitely not this one.

Plus there's no need to split after the CPU when you've got individually controllable branches. I can send as much coolant to the CPU as I please. I've even cranked the pump up to 5 and almost completely shut the GPU and NB branches sending almost all of the flow through the CPU and it didn't increase the OC or decrease the temps noticeably.

These are all really moot points considering your cooling the system with an infinitely better cooling system than air. It's like bickering whether it be more efficient to kill the fly with an RPG or a tank round! Both are overkill and will more than get the job done...
 
Again, I have to disagree about the elbows.

Helped you accomplish what you wanted to do, that's cool. My own preference is to avoid the added restriction they contribute.

In terms of the manifold, do you ever actually throttle the cpu branch? That valve seems redundant. :confused:
 
Helped you accomplish what you wanted to do, that's cool. My own preference is to avoid the added restriction they contribute.

In terms of the manifold, do you ever actually throttle the cpu branch? That valve seems redundant. :confused:

Actually yeah, I used to throttled that branch down when it only had a E6300 on it while the lower branch had (2) hot G80's to cool. Not to mention the 680i on the middle loop!
 
Back
Top