Klingon style 2 pump WC loop setup, need help!

[blackjackel]

I saw this episode of star trek once that said that klingons have two hearts, which is why I call my new 2 pump watercooling loop "klingon style" =P


Aaaanyways, I need help with pump placement, this is how I currently have it set-up:

Pump 1 --> CPU(whitewater block) --> Pump 2 --> Rad/Heatercore --> Reservoir -->pump 1

I believe that this loop setup will give the highest pressure to my waterblock (whitewater), wich is the single most restrictive part in my entire system. Whitewaters are rated to work the best at highest pressure...

this is how I see it:

Pump 1 (PUSHES on to) --> Whitewater --> Pump 2 (PULLS FROM).


The only problem I see is that pump 2's intake will be highly restricted and this might cause the life of the pump to decrease, am I right about this?


Can anyone give me any hints or clues on my pump placement, assuming that the configuration I have isn't the best? What have you guys seen other people do?

[pwnt by pat]

Use whatever gives you the shortest amount of tubing. Don't worry about pump 2, as long as pump 1 is fed directly from the res.

[blackjackel]

Are pumps better at pushing water than at pulling it?

Should I have both pumps pushing on eachother (and then on the waterblock) instead of one pushing one pulling? (example: pump 1 --> pump2 --> waterblock)

[Perseus]

I've got both pumps situated just before my rad. When I had them setup in a test loop with just a rad, no fans, I was amazed at how much they warmed my coolant. As it is now, with a fair amount of tube, my CPU (the first FRU after my rad) idles at 1c above ambient and 4c above at full load.

[pwnt by pat]

You'd probably be better off going res->pump1->pump2 but the pumps ability to "pull" isn't what causes this. The only reasaon to do pump1->pump2 is because you use a res. That's it.

In a loop with a T, even if you have pump->block->block->pump, it won't make any diffference to pump 2

[blackjackel]

You'd probably be better off going res->pump1->pump2 but the pumps ability to "pull" isn't what causes this. The only reasaon to do pump1->pump2 is because you use a res. That's it.

In a loop with a T, even if you have pump->block->block->pump, it won't make any diffference to pump 2

I still don't understand completely, why do you recommend res->Pump1->pump2 over pump1->cpu->pump2 ????

Why JUST because I have a res? What will this configuration change over the one I suggested? What are the advantages? Please elaborate.....






By the way, if your way is the best way to go, I would assume that this would be the best situation res->pump1->pump2->CPU->res, let me know if this is correct.

[thorilan]

because basic laws of science or nature

maximize your DT

[Perseus]

because basic laws of science or nature

maximize your DT

Well this certainly clears things up. :rolleyes:

[thorilan]

its simple what do oyu have to do to maximize your DT?

if you can explaine that ( which anyone who has been here a while can if they have been paying attention) then you have the answere.

been building cooing systems for a while now so think i have a little knowledge

[blackjackel]

its simple what do oyu have to do to maximize your DT?

if you can explaine that ( which anyone who has been here a while can if they have been paying attention) then you have the answere.

been building cooing systems for a while now so think i have a little knowledge

Too bad you have enough of an elitist attitude to not explain things clearly... or at least post links to where information can be read =/ And I'm not saying this to tick you off, I'm just pointing out that these forums are here for teaching as well as learning.

[Otter]

been building cooing systems for a while now so think i have a little knowledge
I'm sure you do, Thorilan. But knowing you know doesn't help us much. I've never had a system with a res, and hence I've never had a chance to test any of the many theories about res and pump placement I've read on this forum. Some of them are obvious nonsense. For instance, one member claimed that with proper res configuration and placement, more water flowed into the pump than out. Amazingly enough, that one went unchallenged in the thread. But some of these ideas may have merit. As you say, you've built quite a few WC systems, and you've probably tried more configurations than most of us and can speak from actual expereince.

[blackjackel]

I'm waiting for soeme responses from some seasoned veterans of watercooling to answer a few of my questions.

The question I want answered the most is:
What is stronger in pumps, the pushing side, or the pulling side....

I'm guessing pushing side, but not sure =./

[Otter]

Centrifugal pumps push better than they pull. They do provide some suction, but if the pressure at the inlet drops too low, a centrifugal pump will cavitate (turn the water to vapor). Because centrifugal pumps are useless for pumping gasses, this is bad.

That said, I'm not sure that in a closed system it matters all that much. The pressure at the inlet of the pump will depend on the flow rate, which should be about the same no matter what the order of the components, except perhaps if you put a shallow, air-filled res immediately after the pump. As I said earlier, though, I have no experience with reservoirs.

[thorilan]

ok since you are un willing to learn for yourself the most important concept in cooling ( DT) i will quote a post i did a long time ago:

you want to cool your most important part the most correct?
that would be the CPU
so since the CPU is the most important part to cool it should recieve the coolest fluid correct?
well when is the fluid in the cooling loop the coldest? that would be right after the radiator correct?

well why is that so important. . to maximize your your DeltaT or dT for short in this post

well what is the delta t you may ask. let me illistrate

say you have a hot frying pan of a stove and your job is to cool it as cool as you can get it .

so you want to pour water on it right?
well do you think pouring hot water on it will cool it as much as pouring ice water? of coarse not .
so you use the coldest water passsible which is ice water.
basically this describes a range.
say the pan is VERY hot and the ice water is very cold so the difference between the 2 is very high ( literally more than 100c ) this is what you call maximizing your dT . you make the seperation between the hot and cold as wide as possible.

this is also why so many noobies get yelled at for preaching " water going slowly picks up more heat"

why is that statement false?
simple
when i pour my ice water onto the hot pan it will quickly become hot water right? so the dT i described earlyer is not so wide any more after pouring ,correct?
so what do you do?
simple
you pour more ice water in because it cools better than boiling water doesnt it?

this also leads to the answer about flow and KEEPING your DT AS WIDE AS POSSIBLE


basicaly the faster your pour the ice water onto the pan the cooler it will stay because ice is very cold and pan is very hot correct?

thats why people many times use the phrase flow>all

if the water just sits there it heats up and eventualy will not cool the pan at all.

now how this applies to cooling computers

you want your water block on your CPU to recieve the coldest possible water . that means putting it right after the rad because if you put anything in between it like say a NB or VGA block you are adding heat to the cool water before it hits your most important part so you are lowering your dT or reducing the distance between the 2 ( cpu temp and coollant temp)

so here is the kicker

your radiator works the exact same way your water block works only in reverse

so you want the hottest possible coolant from the system to hit the radiator last where the radiator is coolest. BING there is the secret or rather the basics of water cooling for building a system based on good solid principles and understanding

[thorilan]

ok so now that you see how the DT works applie that to oyur multi pump configuration

to define your configuration you have 2 pumps which is 2 heating elements

you also have your block. well putting them all in a row before hitting your rad will heat the water more than if you have the radiator between them . but since you want your first heating element to recieve the coolest water to start off that means your block is the starting position in the loop

so your block heats the water first ( this should always be like this unless oyu have an extremely unconventional setup) then both pumps should add thier heat to bring the water to its maximum temp before it hits the radiator

as for pushing and pulling of the pump as long as you dont cavitate the intake you are ok. this means its ok to put SIMILAR rated pumps back to back

if you try putting 2 completely different rated pumps together you will burn out 1 or both pumps

[blackjackel]

ok so now that you see how the DT works applie that to oyur multi pump configuration

to define your configuration you have 2 pumps which is 2 heating elements

you also have your block. well putting them all in a row before hitting your rad will heat the water more than if you have the radiator between them . but since you want your first heating element to recieve the coolest water to start off that means your block is the starting position in the loop

so your block heats the water first ( this should always be like this unless oyu have an extremely unconventional setup) then both pumps should add thier heat to bring the water to its maximum temp before it hits the radiator

as for pushing and pulling of the pump as long as you dont cavitate the intake you are ok. this means its ok to put SIMILAR rated pumps back to back

if you try putting 2 completely different rated pumps together you will burn out 1 or both pumps


I thank you for your response, I read both posts.... I would like to point out that I actually tried to look up what you were talking about, but DT was just too short of a search term... typing "dual pumps" or "pump1 pump2" into the search didn't help either, so I thank you for your response.

According to what you said, you say the best dual pump configuration is:

Reservoir -> Pump1 -> Pump2-> Radiator -> CPU -> Reservoir

That way, the CPU recieves the "coldest" water. There's just one thing to think about, my waterblock is the WhiteWater block, which is one of the most restrictive blocks in the market, and that block operates the BEST at the highest pressure, which is why I suggest either of the following:

Reservoir -> Pump1 -> Pump2 -> CPU -> Radiator -> Reservoir
(I think this will give me the most pressure on the CPU nozzle)

or my original plan

Radiator-> Reservoir -> Pump1 -> CPU -> Pump2 -> Radiator
(not as much pressure on the CPU nozzel, but only 1 pump's heat is dumped on the CPU, not two pumps like listed above)

This would increase the flow THE MOST at the VITAL (and most restricted) part of my system where I NEED the MOST flow.

I believe that the pressure increase that I get on my whitewater by using one of my two configurations will cancel out the heat that the pumps add to the water, giving me better temps in the end.


NOW, based on the information that I provided above, what do you think I should do? Which configuration should I use from the three listed in this reply?

[pwnt by pat]

Flow will remain the same throughout the entire loop
NET pressure will remain the same throughout the loop

[Perseus]

"DT" I think means Delta T, or difference in temperatures. I don't know if you missed this, but I >did< say that I'm using two pumps in series.

Pump1->Pump2->rad->CPU->GPU1->GPU2

With a Storm, also restrictive, I get slightly above ambient (1c) idle, and 3c above during extended load. This would seem to mean that having a rad as the last component before a restrictive block doesn't make a whole lot of difference. That's unless you wouldn't be satisfied with the temps I'm getting and are expecting your operating temps to be lower with your proposed configuration.

Now, knowing this, if you decide to feed your block with water directly from your pumps, you should also know that your temps ARE going to be a bit warmer. I THINK that's what thorilan was trying to tell you.

[Flip-Mode]

Wasn't it proved a while ago that component placement didn't matter? I was sure it did but I'll take your experience as a credential.

[Stratus_ss]

according to This (http://www.overclockers.com/articles1088/) placement does not really matter that much. UNLESS you have some high heat outputs

this guy's basic theory is that for every ~261W another 1 degree C is added to the loop

he says
"So, do what allows for the simplest tubing runs - tubing length/kinking will have a greater impact on temps."

although of course Thorilan is absolutely correct for maximum cooling. I would think that his explaination would be more useful if you had say a BIX 3 that would cool the water alot more effective

my own opinion is that it really depends on the cooling capacity of your rad. The better it cools the more Thorilan's theory holds true

not that I have anywhere near the expertise of Thorilan... nor would I claim to

[blackjackel]

Wasn't it proved a while ago that component placement didn't matter? I was sure it did but I'll take your experience as a credential.

Thanks for reminding me, I do believe I remember what you're talking about, but I don't think it applies to highly restrictive setups like the whitewater block, the pressure before the block will be much higher than the pressure after the block...

[MVC]

The one thing Thorilan didn't mention, however, is that every closed loop will eventually reach equilibrium (say, after about 20 minutes of folding) and that no matter how much you arrange parts to increase ΔT the difference between the "coldest" and "hottest" parts of a "normal" loop in a PC will be no more than 2° or 3°C after equilibrium has been reached. Designing a system to maximize ΔT works best in a system where you can get much larger differences in temperature between the heating component and the cooling component and a PC just isn't big enough for that to happen. (Mostly. You could use monster fans to increase the cooling effect, but you'd then have to live with the noise or you could put the radiator in the basement where ambient is cooler and then deal with long tubing runs--but, everything has to balance out. As long as all of your components are in/on/next-to the case the system as a whole just isn't big enough for there to be great differences between the hottest and coldest components.)

In a loop, inside a computer, what you really want is equilibrium to be reached at the lowest possible temperature--the system stable at 38°C rather than 45°C, for example. And, the easiest way to do this is to have the largest (most efficient) radiator possible, keeping the loop as short as possible, while maintaining the highest possible pressure so that the water keeps moving. Think about it, if you've got a pump rated at 6L per minute and your loop has 1L of water, that water is making 6 circuits per minute. In no single circuit is the water being significantly heated or cooled--but the system as a whole is working toward equilibrium.

I personally like having the pumps in series, one right after the other, because that configuration allows them to maintain the highest pressure throughout the loop (reservoir or no reservoir it's still a closed system and the same pressure will be maintained). But, you could put the pumps as far from each other as possible, and I don't think you'd see a 1°C difference. Plumb the system so that you have nice clean bends without kinking while using as little tubing as possible and no matter what order you put the compenents in you'll be within ~2°C of optimum for those components, in that loop, in that case. A PC really just isn't big enough for component order to make a significant practical difference.

[Perseus]

Don't take this the wrong way jackel, but if you've already made up your mind on how to plumb your loop, and won't even consider contrary opinions\experience, why keep asking for advice?

[MVC]

Thanks for reminding me, I do believe I remember what you're talking about, but I don't think it applies to highly restrictive setups like the whitewater block, the pressure before the block will be much higher than the pressure after the block...
It doesn't matter. You're looking at individual components, but what you need to look at is the system as a whole. No matter how much of a pressure drop you get because of the restrictive block, all the water will still make it back to the pump to make the circuit again. Two pumps in series can maintain higher pressure against that restriction, but for practical purposes even seperating the pumps won't make that much of a difference to the system.

[blackjackel]

Don't take this the wrong way jackel, but if you've already made up your mind on how to plumb your loop, and won't even consider contrary opinions\experience, why keep asking for advice?

No offence taken since no offence was intended, but I almost never completely make up my mind, even when I say that I do, and I don't believe that I've ever said that I've made up my mind in this thread... maybe you just interpreted something I said to mean that?

The whole point in this entire thread is to take several inputs, compute them all in my head, and ultimately decide (based on all input given) among with other inputs to make the final decision of how I'm going to set up the rest of my system....

[blackjackel]

It doesn't matter. You're looking at individual components, but what you need to look at is the system as a whole. No matter how much of a pressure drop you get because of the restrictive block, all the water will still make it back to the pump to make the circuit again. Two pumps in series can maintain higher pressure against that restriction, but for practical purposes even seperating the pumps won't make that much of a difference to the system.

This is perhaps the most eye-opening response I have gotten so far....

I called the the owner of Dtek customs and have been told by him that hooking up two pumps side-by-side might not be a good idea.... he told me something that I can't remember, but what I got out of it is that one pump will partially cancel the effect of the other pump somehow... sucks huh?

I've learned so much so far from this post, I've re-arranged my tubing and filled and bled my system at least 5 times so far and have done lots of testing. I'll post my final results with possible pictures of my final result....

[thorilan]

the only time you want to run pumps in parrallel is if the pumps have such good head and the system overall resistance is so low that you benifit from a lot more flow because there is no resistance to overcome.

the only times i have sen this situation is when i personally have designed vidoe loops . cpu loops are just to restrictive .

putting 2 pumps next to each other is called parrallel and moves twice as much water at the same head rating . the problem is instead of moving water a lot faster the amount of restance in the system often slows the flow down from them being parrallel more than if they where in series.

that is why most people run series .

in real world tests done at a .2c resolution with diferent tubing and order configs i found in 87% of the loops that a pump at the bottom of the loop with them configured the way i told you showed anywhere from .6c to 3.2c diference with an average 2.2c better than any other config.
keep in mind almost all the loops where with single pass rads though so the tubing runs are naturally shorter in the correct config anyway.

[blackjackel]

the only time you want to run pumps in parrallel is if the pumps have such good head and the system overall resistance is so low that you benifit from a lot more flow because there is no resistance to overcome.

the only times i have sen this situation is when i personally have designed vidoe loops . cpu loops are just to restrictive .

putting 2 pumps next to each other is called parrallel and moves twice as much water at the same head rating . the problem is instead of moving water a lot faster the amount of restance in the system often slows the flow down from them being parrallel more than if they where in series.

that is why most people run series .

in real world tests done at a .2c resolution with diferent tubing and order configs i found in 87% of the loops that a pump at the bottom of the loop with them configured the way i told you showed anywhere from .6c to 3.2c diference with an average 2.2c better than any other config.
keep in mind almost all the loops where with single pass rads though so the tubing runs are naturally shorter in the correct config anyway.

Thanks for the input, but what exactly is series? is that when you have two pumps with something in between them? Eg: pump1 -> SOMETHING ->pump2?


I now completely agree with you that I must have at least one pump with the intake from the reservoir, this will help IMMENSELY with bleeding the system. See, you just have to give me a few details of WHY a certain way is better... not just say it's better =)

From what you said, I understand that you recommend the pumps be run in seris, not in paralell (because of my higly restricted waterblock). Since you sound like you have alot of experience, I'll listen to you (and that other guy that sounds like he knows what he's talking about) the most.


So far, this is what I'm SURE about:

Reservoir-> pump1

And this is what I THINK that I should do (based on the information that I've been getting that recommends me running the pumps in series rather than paralell)

Reservoir -> pump1 -> Radiator -> Pump2 -> CPU -> Reservoir

OR

Reservoir -> Pump1-> Radiator-> CPU -> Pump2 -> Reservoir

Do you recommend I do one of the above, or do you have your own layout that you think I should do? If you have your own layout you recommend, please let me know WHY you recommend it.

[Otter]

Thanks for the input, but what exactly is series? is that when you have two pumps with something in between them? Eg: pump1 -> SOMETHING ->pump2?
Series means "one after the other". Pump>Something>Pump and Pump>Pump>Something are both series configurations.

[blackjackel]

Series means "one after the other". Pump>Something>Pump and Pump>Pump>Something are both series configurations.

Thanks for the explanation, but I'm still kinda confused about what I should do about the pump setup, I have a picture, but it's not 100% clear yet...

[blackjackel]

because basic laws of science or nature

maximize your DT

According to this comment, I am assuming that you agree with Perseus in that the pumps should be reservoir => Pump1 => Pump2

But do you also agree with that other peron's recommendation that both pumps should be just before the radiator and cpu, like this:

reservoir => Pump1 => Pump2 => Radiator => CPU => reservoir

????

[Otter]

I suggest res>pump>pump>shortest route through the rest of the components avoiding tight bends. That way, you can have both pumps at the bottom of your case (the extra water pressure helps prevent cavitation), you'll have easy bleeding, and you'll keep restriction and clutter to a minimum. With two pumps, your flow rate is going to be high enough that putting the CPU after the rad won't give a significantly higher delta T. But, if it doesn't require more tubing or bends, the ideal would be res>pump>pump>rad>CPU>GPU>res. I think that having a short, clean route is going to do more for your performance than having the rad right before the CPU, but keep those pumps on the bottom like Thorilan suggested.

Oh, and whatever you do, avoid tight bends right at the inlets of the pumps.

[blackjackel]

I suggest res>pump>pump>shortest route through the rest of the components avoiding tight bends. That way, you can have both pumps at the bottom of your case (the extra water pressure helps prevent cavitation), you'll have easy bleeding, and you'll keep restriction and clutter to a minimum. With two pumps, your flow rate is going to be high enough that putting the CPU after the rad won't give a significantly higher delta T. But, if it doesn't require more tubing or bends, the ideal would be res>pump>pump>rad>CPU>GPU>res. I think that having a short, clean route is going to do more for your performance than having the rad right before the CPU, but keep those pumps on the bottom like Thorilan suggested.

Oh, and whatever you do, avoid tight bends right at the inlets of the pumps.


Thanks for the recommendations and such, I think I'll do as you suggest, but what about those who have told me that putting a pump RIGHT after the other pump will be detrimental to performance? What do you have to say about that?

[jamesavery22]

...snip
in real world tests done at a .2c resolution with diferent tubing and order configs i found in 87% of the loops that a pump at the bottom of the loop with them configured the way i told you showed anywhere from .6c to 3.2c diference with an average 2.2c better than any other config.
keep in mind almost all the loops where with single pass rads though so the tubing runs are naturally shorter in the correct config anyway.


Got any links thorilan? A loop that gives the least restriction, as you said they were shorter anyways, has more affect than placing the rad between the pumps...

And can you give a little more info on the loops? Maybe you were sleepy when you posted or something as Im sure you know just giving water temp differences is useless without everything else...

[pwnt by pat]

Thanks for the recommendations and such, I think I'll do as you suggest, but what about those who have told me that putting a pump RIGHT after the other pump will be detrimental to performance? What do you have to say about that?

I don't believe anyone said putting the pumps one after another would be detrimental to performance. It should make the loop overall more efficient than running pump-block-pump but only really because the overall net pressure might be reduced as the second pump will have to use some of it's energy to "suck". The only performance loss "thing" would be pumping directly into the resevoir which may or may not have any affect on performance. NPSH

[blackjackel]

I don't believe anyone said putting the pumps one after another would be detrimental to performance. It should make the loop overall more efficient than running pump-block-pump but only really because the overall net pressure might be reduced as the second pump will have to use some of it's energy to "suck". The only performance loss "thing" would be pumping directly into the resevoir which may or may not have any affect on performance. NPSH

Input taken, just a quick question, do you agree with otter's suggestion?

I'm just wondering what the general concencus is, because now I'm thinking of doing this (otters suggestion) after all input taken:

reservoir -> pump1 -> pump2 -> CPU -> heatercore -> reservoir

[pwnt by pat]

My only input is res->pump->pump-> shortest amount of tubing

[thorilan]

Got any links thorilan? A loop that gives the least restriction, as you said they were shorter anyways, has more affect than placing the rad between the pumps...

And can you give a little more info on the loops? Maybe you were sleepy when you posted or something as Im sure you know just giving water temp differences is useless without everything else...
nope these are from my own literal thousands of tests run on over a hundred systems

and as for what the loops look like . they are all custom to customer needs. i have cooled everything from minimultis to sound editting server in studios and even some stuff im under either nda or secrecy agreements

[Otter]

Thanks for the recommendations and such, I think I'll do as you suggest, but what about those who have told me that putting a pump RIGHT after the other pump will be detrimental to performance? What do you have to say about that?
Not much. I've never heard that before and don't now why it would be true. But if you post the theory here, I'll give it a think.

[learning curve]

I read somewhere Cathar says that you can run 2 mismatched pumps, but for best performance the weaker should feed the stronger of the 2. Hope this helps a bit. Good luck with your setup.

[thorilan]

you will burn out 1 or both pumps doing so. been there done that