Water cooling loop order question.

  • Thread starter Thread starter Gdog96
  • 17 comments
  • 4,465 views
G

Gdog96

Messages
766
United Kingdom
England
I have heard a lot of people say that as long as the pump goes after the reservoir the loop order doesn't matter.

I just want to verify something.
I can go from the res to the pump and then to all of my component blocks before even touching a radiator?

Because I was thinking, surely on startup the CPU would get really hot by the time the cooled water reaches it as the water has to effectively go round the whole loop twice to reach the CPU after being cooled by my rad.

This may sound like a stupid question but I want to make sure before I undertake a watercooling project.
 
Last edited:
Mine goes res, pump, rad, cpu, gpus, and back to res. As you said it doesn't matter if the rad is at the beginning or end of the loop. The water is plenty cool on startup so you'll be fine.
 
Ok thanks for clearing that up for me David. I'll make sure to post some pictures on here when my watercooling is done in probably about 2 weeks. I've already ordered everything :D
 
David
Mine goes res, pump, rad, cpu, gpus, and back to res. As you said it doesn't matter if the rad is at the beginning or end of the loop. The water is plenty cool on startup so you'll be fine.
Click to expand...

I would link you would want to return cold water to the res, not hot water.

@Gdog96,
Pretty standard order would be res>pump>components>rad.

As far as component ordering, if your water flow is good enough, the water temp should equalize throughout your loop with a stepping increase of temps as you stack more components. I think it's a good choice to go to the CPU first as most CPU blocks offer really good water flow.
 
Pako
I would link you would want to return cold water to the res, not hot water.

@Gdog96,
Pretty standard order would be res>pump>components>rad.

As far as component ordering, if your water flow is good enough, the water temp should equalize throughout your loop with a stepping increase of temps as you stack more components. I think it's a good choice to go to the CPU first as most CPU blocks offer really good water flow.
Click to expand...

I just have a Swiftech micro-res so the water is in and out of the res pretty quickly. I think whatever I had read when I was putting everything together suggested that order so I went with it :lol: Had never originally planned for water so next time I'll get a better case for it and arrange everything more optimally. At any rate, my temps have never been an issue.
 
No worries, if at the end of the day it works then you don't need to change a thing. 👍
 
Thanks guys, I don't suppose you guys could answer this too as it doesn't deserve it's own thread. I want to know if my GPU has a reference design PCB because if it does that means I can fit a water block to it, as I'm not aware of any water block that fits a non reference design. I have the Gigabyte GTX 660 non Ti version. I can't find annything on whether it's got a reference design or not. Do you guys know or could point me anywhere where I could find out? Thanks again.
 
From what I've read it seems there really isn't any reference design for the 660's. You can probably find a generic water block that will fit on the GPU and match up with the holes. Looks like most 660s are about the same though and Gigabytes doesn't look unusual so you're probably safe.

Found this to start for a Gigabyte 660.

http://www.coolingconfigurator.com/step1_complist?gpu_gpus=865
 
Last edited:
Best way if doing a CPU cooling would be
Res>Pump>Rad>CPU block>Res.
But I have seen it go
Res>Pump>CPU>Rad>Res.

If doing a Multi component cool
It would be
Res>Pump>rad>CPU>GPU>res.

Best way to do a multi cool loop is
Res>Pump>Rad1>CPU>Rad2>GPU>Res

This way the coolant stays cool for each device.
 
I was just searching the internet when I stumbled upon this thread. I had to make an account to post here. The order makes no difference. Radiator to cpu then gpu the res. Or res to components to rad back to reservoir. There isn't a HOT water cool water. The entire loop stays the same temperature and there are plenty of tests to support this. One test I know put a sensor at the reservoir after 2 radiators back to back then ran the loop to CPU and 4 GPU's then measured the temp there. The 2 sensors had a delta of about .6C. So run what will give you the least amount of tubing required and be sure that you atleast have enough radiator for what you are running. Its always better to be on the too much side than on the too little side. GL on your build and enjoy the benefits of water cooling.
 
FunktasticLucky
I was just searching the internet when I stumbled upon this thread. I had to make an account to post here. The order makes no difference. Radiator to cpu then gpu the res. Or res to components to rad back to reservoir. There isn't a HOT water cool water. The entire loop stays the same temperature and there are plenty of tests to support this. One test I know put a sensor at the reservoir after 2 radiators back to back then ran the loop to CPU and 4 GPU's then measured the temp there. The 2 sensors had a delta of about .6C. So run what will give you the least amount of tubing required and be sure that you atleast have enough radiator for what you are running. Its always better to be on the too much side than on the too little side. GL on your build and enjoy the benefits of water cooling.
Click to expand...

First off, thank you FunktasticLucky for taking the time to register to provide your input. Clearly this is a subject you are passionate about! Secondly, welcome to GTPlanet. We share many passions here and are a great community.

In response to your post, I would say that there are just too many variables in the "plenty of tests" to say that the entire loop stays the same temperature. I would agree that given the right environment, the loop would be the same temp throughout. This could be a good thing, or not a good thing, the temperature of the loop would determine that. There are some contributing factors that will maintain a constant temperature loop. 1.) High water flow 2.) Short loops 3.) Small or no reservoir.

I only say the above because I have plenty of personal tests that tell me I have a cold side and a hot side. I have taken the liberty to take some extreme measures in my loop to assure I have plenty of fresh and cool water/coolant available, I have also taken measures that keep my water flow at a minimum and have taken some pretty extreme measures to make sure my radiator is as effective as it could be without designing a cooler.

If you're water flow is too fast, you are not allowing adequate time for heat transfer to occur, both at the blocks and at the radiator.

Again I will say, with smaller loops with a more limited cooling design, equalization will occur in the loop. You can, however, achieve superior cooling if you can adjust some of the variables mentioned above.

:cheers:
 
That is the misconception. The big thing you need is turbulence. Higher flow will increase turbulence which causes the molecules to mix with other molecules and increase the amount of heat absorbed by a unit of water.

Think of it this way... on a hot day what cools you better? A fan on low speed or a fan on High speed? This doesn't change whether it be air or water. Teh higher flow of water will increase the effeciency of the loop and run cooler temperatures to a certain extent. You can have diminishing returns where your pump will be running such high flow rates that it dumps more heat into the water than the increased flow benefits. Skinneelabs has done all the testing for you though if you don't believe me.

A much simpllified way to look at it is this... the Water loop is a closed loop system. Meaning everything has to be treated as a reservoir. The faster the water moved the more times it completes a circuit and goes through the radiator. The more times it goes through the radiator the cooler the water is. The cooler the water is the higher the Termperature delta between the Block and the water. And we all know that heat transfer effeciency increases with a higher temperature delta. The main thing to really focus on is the head pressure of the pump. If you can keep total flow around 1.5GPM you're in the best area. But if you start slowing the rate down you will increase temps drastically and it's an exponential function so as you start slowing down the hotter it gets and DRASTICALLY.

Here's a link to show you some tables. http://skinneelabs.com/i7-blocks-2/2/

so
 
I guess I better get a bigger pump or a second pump for my system then. :) at 0 ft of head I am pushing >700GPH, I have total rise of about 2 feet total in system but have a rather large loop and find the flow to be minimal. At first I thought this was a concern, but under load, my temps are very good. With what you are saying, I should achieve even better results with more flow. I guess I better find me a good inline pump and start testing. Not to hijack this thread, but you can see my build here.

I know without a cooler, there is no way to maintain ambient temps, but I wanted to give it my best effort.

:cheers:
 
Pako that is one of the most interesting things I have seen sin the dude who built his PC inside a fish tank with fish in it. The PC was separated of course but was interesting.

May I suggest if you are game and haven't done so already, delid your ivy bridge and apply your own thermal paste. The crap intel used is trash and peoplw have seen 27C drops at load :o. My 3770 will hopefully be arriving this week along with my 2nd 680. Also, instead of the box fan for that radiator I would be curious to see the cooling ability if you used a car radiator fan. Doubt it would make much difference as that thing has ridiculous surface area and its cooling ability far surpasses the heat generated.

But going back to what I was saying earlier. With the flow moving quickly the closed loop will reach equilibrium and stabalize. The temp of the water may vary less than 1C in places which is negligible. Therefor the order in which you run the components does not matter. The key is to make the loop as short and least restrictive as possible. The fewer kinks and 90 degree fittings used the more effecient the loop will be. I'll be building my loop as soon as the obsidian 900d is launched. I plan on a 480mm and 360mm radiator. Should be pretty sufficient for a 3770 and 2 680's.
 
LOL, yes...it has been an interesting build for sure.

I had a bad experience back in the day with a Voodoo 5500 and a razor blade. My fine motor skills just aren't what they used to be. :D As far as a radiator fan, believe me, I looked. Prices were way to high and were all 12v motors as well. That box fan cost $17.49 I believe at Home Depot. Cost of the plexy was about $12. Those radiator fans started around $140 if memory serves me correctly. What I would benefit from is the coned baffle that would taper down to the radiator. I have entertained a few designs and still may pursue that down the road.

Again, I agree given enough flow, your loop will equalize. The larger the reservoir, the more flow is required to reach that equilibrium. Granted my setup is not "normal" in the traditional sense, but the larger tank and having 8.5-9 gallons of cool water ready to feed the loop, I am assured colder water going in than I have coming out of the components. I have looked into some Little Giant pumps but I honestly haven't had any issues with the current setup and I have left it alone.
 
Oh man. I miss 3dfx :( I loved them from the moment my brother installed a voodoo2 12MB card in his pentium 133 haha. Then we got a pentium 2 300MHz and I went from a diamond voodoo banshee to a 3dfx voodoo3 to a voodoo5 5500. Then nvidia bought them and shut em down. Man I feel old haha.

I bet you could find a fan in a junk yard really cheap. Go for a lincoln mark 8 fan. Highest flowing fan +n a vehicle. You'll have to buy a transformer though. 12v is nothing. But the amperage on them is through the roof haha. It was just an idea though.

I have more money than sense so I will be taking a stab at poppin the top off my ivy bridge. Haha wish me luck.I hopefully won't accidently cut the PCB or the chip itself.
 
LOL, I do wish you luck. Looks like 74% have success poppin the top off according to Overclockers.net poll. Fairly decent odds. :)
 

Latest Posts

Back