Watercooling Build. Dual loop thoughts.

[SeeThruHead]

Is there a formula to determine flow rate given pump psi and component pressure drop? or will i need more information? pump psi and pressure drop of the radiators i can find on skinneelabs but i dont see those listed for water blocks or fittings/tubing. Although i've read pressure drop for fittings and tubing are really quite small. Also, when determining how much many watts of heat and overclocks system will produce, do you use the processor TDP values? Or will wattage go beyond that when overclocking to the extremes?

 

[Conumdrum]

It's out the window on a rig like this. Martins flow rate esimator is old and has no data on better (Bitspower) fittings etc. Your kinda on your own.

Heatload is talked about in the stickies. You can kinda figure out your wattage, but it's not a science. It's a hobby. Overclocking to extremes isn't what watercooling is about. It's good above normal air clocks and quiet. Extreme is another step up the food chain. Phase etc.

Build it, your not gonna break anything. Will it be perfect? Dunno, you'll adjust as needed. It's a hobby thank goodness.

I would NEVER build such a rig for a starter, I don't have the $$ or the need.

Your building a rig that only a few build, and most if not all of them have years of experiance in watercooling.

Your overraded, add fans as needed. You'll be fine as long as you have no leaks etc. I can just imagine the complexity of the annual teardown and cleaning every teeny bit. My rig is enough pain, and started my redo as I type.

 

[utnorris]

So just so I am clear, you will be using 4 rads to cool just a SB chip? If this is correct, the 4th rad actually may work against you, even if it's only slightly. I have the 2600K overclocked to 5Ghz with 1.44v and two GTX460's all in the same loop. I am using 3 x Thermochill triple rads on the front with 9 x Feser Triberk fans, my water temp stays at 2-3 degrees above the ambient air intake. If you put the rad on the back, then yes, you will be running warmer air across it than the ones on the front, which would hurt your temps, even if it is ever so slightly.

Conumdrum is correct, there use to be an estimator for flow, but it's way outdated. I am not sure if you have done this, but here is a link to a bunch of info, granted it may have some of the info on this site duplicated, but it might be good for you to take some time and read a few. They may be able to answer your questions with a little more detail.

http://www.xtremesystems.org/forums/showthread.php?t=202394

 

[SeeThruHead]

Yeah I guess I'm kinda a go big or go home mindset in a lot of things. (Just got my first tattoo - full sleeve) I'm fine with that. Ill be able to run a cpu and 2 vid cards with a fairly low deltaT with fans at low speed. (1450) sounds to me that's exactly what watercooling is all about. Also, would hot air from the three front rad's actually be able to HEAT up the water when passing through the rear radiator? That's something I hadn't even thought of. And if thats not possible then perhaps the amount of hot air going through the exhaust rad will make it so that it doesn't provide any benefit at all. I could however change that rad to intake, and use some higher cfm fans for the remaining 5 unrestricted fan holes. HRM...
EDIT: I think I'll just settle on 3 rads in intake with 7 unobstructed holes for exhaust. Saves a good deal of money and I'm overradded as it is. Man I can't wait to get this build started!

 

[m0r7if3r]

So just so I am clear, you will be using 4 rads to cool just a SB chip? If this is correct, the 4th rad actually may work against you, even if it's only slightly. I have the 2600K overclocked to 5Ghz with 1.44v and two GTX460's all in the same loop. I am using 3 x Thermochill triple rads on the front with 9 x Feser Triberk fans, my water temp stays at 2-3 degrees above the ambient air intake. If you put the rad on the back, then yes, you will be running warmer air across it than the ones on the front, which would hurt your temps, even if it is ever so slightly.

Conumdrum is correct, there use to be an estimator for flow, but it's way outdated. I am not sure if you have done this, but here is a link to a bunch of info, granted it may have some of the info on this site duplicated, but it might be good for you to take some time and read a few. They may be able to answer your questions with a little more detail.

http://www.xtremesystems.org/forums/showthread.php?t=202394

This is incorrect. Airflow at all will move heat away from a rad, regardless of the temperature, it's some super complicated thermodynamics and fluid dynamics stuff that I can't really explain (cause I don't understand it), but thorillian explained it once and he knows his ****. The only way an additional rad can hurt (excepting extreme scenarios yada yada yada, don't nitpick etc) is if it decreases the flow such that it drops below some critical point (if I had to eyeball it...idk, .4gpm or so)

 

[utnorris]

Do you have a link? If you are moving warmer air over one of the rads than the other rads, how does this not effect temps in a negative way? This is why stacking rads does not work either regardless of how you flow the water from one rad to the next. Rads can not cool lower than ambient, just cannot, simple physics. Will it have a huge impact? No, but I already stated that. My point was that it wouldn't help the loop, but instead hurt the loop, even if it is ever so slightly. It's already more rad space than needed for either a just a CPU or a CPU and Dual GPU setup. But, as I stated before, whatever the OP wants is fine as long as he/she is happy with the results.

 

[m0r7if3r]

Do you have a link?

Sorry, no, but the phsyics behind it is sound. It's to do with the surface temp of the rad and the velocity of the air and everything...it makes sense once you've seen the explanation (assuming you have enough phsyics background to actually get it...which surprisingly enough most of this forum does), but like I said, it's complex thermo and fluid dynamics and that's not my area of expertise.

 

[hokiealumnus]

Do you have a link? If you are moving warmer air over one of the rads than the other rads, how does this not effect temps in a negative way? This is why stacking rads does not work either regardless of how you flow the water from one rad to the next. Rads can not cool lower than ambient, just cannot, simple physics. Will it have a huge impact? No, but I already stated that. My point was that it wouldn't help the loop, but instead hurt the loop, even if it is ever so slightly. It's already more rad space than needed for either a just a CPU or a CPU and Dual GPU setup. But, as I stated before, whatever the OP wants is fine as long as he/she is happy with the results.

Makes sense.

If you study the diagram in the OP, he's got three rads cooled with ambient air and one cooled by air that went through the other three. They'll put out hotter air than they drag in. If the three of them get delta T low enough, the heat they give off would have a negative affect on the water going through the fourth rad. It sounds logical anyway. I remember Thorilan's post, but am not sure how it applied; as in, whether he took into account warmer air going into the other rad or not. Hopefully he'll happen upon this and enlighten us.

Of course, we should stress the 'ever so slightly' part. He's quite over-radded as it is with three 360's to begin with. SB does not get hot. Running WPrime 1024M at 5.4GHz and 1.6Vcore with my bench loop (single MCR320 & EK Supreme HF), temps barely touched 80°C.

 

[utnorris]

Makes sense.

If you study the diagram in the OP, he's got three rads cooled with ambient air and one cooled by air that went through the other three. They'll put out hotter air than they drag in. If the three of them get delta T low enough, the heat they give off would have a negative affect on the water going through the fourth rad. It sounds logical anyway. I remember Thorilan's post, but am not sure how it applied; as in, whether he took into account warmer air going into the other rad or not. Hopefully he'll happen upon this and enlighten us.

Of course, we should stress the 'ever so slightly' part. He's quite over-radded as it is with three 360's to begin with. SB does not get hot. Running WPrime 1024M at 5.4GHz and 1.6Vcore with my bench loop (single MCR320 & EK Supreme HF), temps barely touched 80°C.

Completely agree here. My 2600k at 5Ghz and 1.44v barely tops 71c with ambients of 29c under full load from WCG. At 4.8Ghz and 1.36v with ambients of 27c under full load of WCG tops out at 63c. This of course is a loop that also includes two GTX460's.

 

[SeeThruHead]

Makes sense.
Of course, we should stress the 'ever so slightly' part. He's quite over-radded as it is with three 360's to begin with. SB does not get hot. Running WPrime 1024M at 5.4GHz and 1.6Vcore with my bench loop (single MCR320 & EK Supreme HF), temps barely touched 80°C.

It seems like it's not going to add MUCH to the cooling potential for the system, and therefore not be worth the cost. So I'm going to skip it for now. I might do what I suggested before in my post and switch it to intake with higher CFM fans on exhaust some day in the future if I need it, but I doubt I will. I'm definitely still over radded but that doesn't bother me one bit.
From what I've read it's possible for these Sandybridges to put out a lot of heat if you enable PLL overvoltage. It also remains to be seen what we'll be able to do with the socket 2011 Sandybridge E chips. Also I want this rig to outlive Sandybridge, and the next processor upgrade and I do after that, same goes for GPU upgrades and overclocks. This setup should be capable of doing a deltaT less than 5 degrees if my load stays under 708 Watts. if at some point in the future I load it above 700 watts I'm still going to have a decent delta which is important for my piece of mind when it comes to future proofing.

 

[hokiealumnus]

FWIW, unless you're shooting for a 50x multiplier or more for 24/7 operation (I wouldn't recommend it for a 32nm chip), you don't need the PLL override. In fact, it gives you a Vcore penalty if enabled below 50x in my experience. For 24/7 I settled on 1.256 Vcore loaded @ 4.6 GHz. It runs really cool and really fast.

 

[SeeThruHead]

FWIW, unless you're shooting for a 50x multiplier or more for 24/7 operation (I wouldn't recommend it for a 32nm chip), you don't need the PLL override. In fact, it gives you a Vcore penalty if enabled below 50x in my experience. For 24/7 I settled on 1.256 Vcore loaded @ 4.6 GHz. It runs really cool and really fast.

From what I've read I'd probably be able to push 5.2 without PLL override, and I could probably do that on air. Buuuut what about 5.4? or higher. Not sure if the chips are capable but the ability to cope with such a load even if I'm not doing so right away is good in my mind. Not to mention 2, 3, or even 4 overclocked GPU's which I've seen can put out upwards of 300watts each.

 

[hokiealumnus]

For 24/7 use, you won't see that kind of speed unless you really don't value the longer term integrity of your chip. The Internal PLL Override is required for any multiplier north of 50x. You won't be able to boot higher without it; the system will just give you a blinking cursor in the upper left hand side of your monitor.

For benchmarking, sure, you can run it that high if your chip will allow you. These are rather fickle beasts, with some topping out at 51x and others topping out at 56-57x. I got lucky and got a 55x chip, but splat's chip tops out at 53x. Both were purchased from the same Newegg warehouse on the same day at the same time. See this thread at HWBot for a good list of batch numbers and where they're topping out. Some batches are better than others, but even within the batches there is a lot of variation. Just saying YMMV north of 50x.

I completely agree with you though, if you're planning on adding that many GPUs to your loop(s), definitely over-rad to start with. Better to have enough than need to add more.

 

[SeeThruHead]

yeah I dunno if i'm actually going to run it that high, but i'd like the option.
As fior your comment on cpu ppl override. i think i was misinterpreting something in this link (or maybe i wasn't)
I think he mentions that his d1 chip doesn't have the ability to enable pll override but he still manages to get to 5.2. Anyway I wont be getting a d1 chip obviously to that point is really moot.

I've run into a really frustrating scenario though (god EK customer service is very quick but also completely useless)
They only sell their EK Multioption res X2 which is incompatible with their DDC dual top....(the x2 is 60mm diameter and the compatible version is 50mm) They offer this but i think that only works to connect one 50mm tube to a 60mm tube which isn't what i need at all.
Edit: nvm, i had to select 3 different res accessories in order to be able to mount a res easily onto the dual ddc top (they should really update this top to accept their new reservoir standard...)
for those who are interested here it is:
DDC DUAL pump top connected to Multioption Res Bottom REV 2 with Rev 2 anticyclone connected to Multioption 50mm diameter tube (not X2) connected to multi option top

 

[SeeThruHead]

Part // Price // Quantity // Total
Laing DDC Rev. 3.25 Liquid Cooling Pump - (12 Volt -18 Watt) // $64.99 // 2 // $129.98
EK-DDC Dual TOP V.2 G1/4 - Black Acetal // $62.07 // 1 // $62.07
EK-Multioption LINK 1×G1/4 // $11.05 // 1 // $11.05
EK-Multioption Tube 140mm // $19.88 // 1 // $19.88
EK-Multioption TOP W/Thread // $9.11 // 1 // $9.11
Danger Den Fillport - Black // $13.99 // 1 // $13.99
Swiftech MCR-320 Quiet Power Series Radiator - Matte Black // $49.95 // 4 // $199.80
Gentle Typhoon 120mm Silent Case Fan 1850RPM // $15.99 // 18 // $287.82
Tygon A-60-G Norprene 1/2" ID (3/4" OD) // $3.00 // 15 // $45.00
Bitspower Black Sparkle G1/4" Barb Fitting - 1/2" ID // $1.00 // 8 // $8.00
Bitspower 1/2" 90 Degree Rotary Compression Fitting G1/4 With BARB // $11.99 // 4 // $47.96
Bitspower 1/2" 45 Degree Rotary Compression Fitting G1/4 // $10.99 // 5 // $54.95
Lamptron Elite Aluminum Reusable 3/4" OD Liquid Tubing Clamps - Anodized Silver // $3.99 // 13 // $51.87
Koolance QDC (Extreme Flow) No-Spill, Female Compression (13mm, 1/2" x 19mm, 3/4")(VL4N-F13-19S) // $15.99 // 1 // $15.99
Koolance QDC (Extreme Flow) No-Spill, Male Compression (13mm, 1/2" x 19mm, 3/4")(VL4N-M13-19S) // $14.99 // 1 // $14.99
Indigo Xtreme™ Engineered Thermal Interface (ETI) Kit for Core i5, LGA 1156 - 2 Pack // $19.99 // 1 // $19.99
Arctic Cooling MX-4 Super High Performance Non-Conductive Thermal Compound (4g) // $11.99 // 1 // $11.99
CPU water Block // $75.00 // 1 // $75.00
GPU water Block Fullcover // $105.00 // 1 // $105.00

Ascension CYO (Choose Your Own)[ASCENSIONCYO] // $529.00 // 1 // $529.00
Lamptron FC Touch - 30W - 6 Channel Aluminum Rheobus w/ Touch Screen - Black (FCT) // $74.99 // 1 // $74.99
Lian Li 5.25" Multi-Media I/O Ports - Black - (BZ-U02B) // $28.99 // 1 // $28.99
Lian Li 5.25" to 4 x 2.5" HDD Converter - Black (BZ-525B) // $24.99 // 1 // $24.99
Lian Li EX-H22B SATA Hot Swap HD Cage - Dual 3.5" Drives - Black // $64.95 // 1 // $64.95
MDPC SLEEVE-KIT // $50.00 // 2 // $100.00
Accessories - Nexus Anti-Vibration PC Mounting Kit - 60 components // $14.95 // 1 // $14.95

TOTAL: $2,022.31

 

[SeeThruHead]

I came across this while lurking another OC board.
http://martinsliquidlab.org
They have a spreadsheet on there that actually looks pretty up do date and useful...

 

[Conumdrum]

Wow, didn't know he redid his old site. His old site was on Skinnee's site, but it's good to see him back. He 'retired' for over a year. He 'was' the king of watercooling science for a few years. Good to see him back. Thanks!

I'll take a look at the flow rate checker later, much updated from his old one.

 

[m0r7if3r]

Skinnee's actually planning on redoing all of martin's old testing that's still relevant

 

[SeeThruHead]

Well it seems skinnee and martin are working together to some extent or at least sharing info. A lot maybe even most of the numbers in the flow spreadsheet on that site are actually skinnee's data. each set of data has a tag showing where it came from [SL] or [MLL] so you can choose who's data you use. I used it to guestimate (becuase as stated in the spreadsheet itself the error can be upwards of 20%) the flow rate for my setup. Just to see if it's within an acceptable range. which it is. anyway. anyway, it's just another source of info from a person who seems to be somewhat renowned in the community.

 

[Martinm210]

Howdy guys...
Glad the estimator is still of some use. I'll try to keep it updated with new tests as they are done. Skinnee sent me over a bunch of his data, so it is a compilation of sources. That ok though, there is plenty of error in sample variance as well, so precision is pretty much out the window unless you wanted to calibrate meters professionally and test a much larger sample size. I've just tried to include the [source] in brackets.

Hope it's helpful and if nothing else a good learning tool.