Pump pulling from CPU does it make a difference?

[bluehaze]

OP, as for foam, did you clean out the radiator?
i noticed some foaming when i used some coolants such as the swiftech and primochills... but in distilled, no fizz... no foam...

let it run a while and see if it clears.
i too have read somewhere that the temp in a loop pretty much normalizes...

I didn't clean it. Was going to take the Rad off and flush it before filling but totally forgot to do it. However now after running for a few hours there is no more foam and the water is clear not cloudy anymore. I was using Primochill Pure so maybe that's why the foam was there or maybe it was from not cleaning it first but everything is crystal clear now so it should be fine right?

[bluehaze]

nah eric and dave... sadius is correct when it comes with the KL-350.

trust me ive had too much time on the bed with it.

Pump right b4 kl-350 on this block guys.

How much difference does it make? Wish I saw that before filling it and bleeding it LOL I suppose I will just leave it as is now unless the difference is really significant?

[nikhsub1]

It will have the exact same pressure drop no matter where it is in a loop.

If a pump produces 100 units of pressure (at the flowrate of the given loop), the rad has a pressure drop (at this flowrate) of 20 units, the res has a pressure drop of 5, and the CPU block a pressure drop of 75, it will be 75 no matter where it is in the loop. X units of pressure will exist at the intake and there will be X-75 units immediately after the block.

Doesn't matter if the order is 75-20-5, 20-75-5, 5-20-75, or whatever. The pressure drop of any given component is a given curve like these:
That curve doesn't change depending on where it is in a loop. Therefore, the pressure drop doesn't change depending on where it is in a loop, therefore the flow doesn't change, therefore the performance doesn't change.

Pressure throughout a loop DOES vary depending on the position in the loop. Remember, pressure is a cumulative beast and each component will have a different PD depending on flow and where it is in the loop. Your above statement is over simplified and not correct. Pressure drop will be different with different amounts of pumping power. The more pump, the more PD we experience. When you think of components in a loop and pressure drop you have to look at the loop as having an absolute beginning and end. The beginning and the end are at the pump outlet (beginning) and the pump intake (end). In your example of 75-20-5 and 20-75-5 etc it is wrong to think that no matter the loop order that each component will RECEIVE the same pressure as it would elsewhere in the loop. It does not work that way. The way to look at your example is in percentages and NOT absolutes as you have presented them. If the WB is right after the pump, it will take 75% of the pressure of the pump. This leaves 25% left for the rest of the loop. If the RAD were to be first as in your example, it would take 25% of the total pressure and leave 75% of it to the rest of the loop, so if the CPU block were next, it would only get 75% of the pressure...

A component's (block or radiator) thermal performance directly relates to flow and flow alone.

Yes, but flow is relational to pressure squared.

But the pressure drop of a component is constant no matter where it is in a loop (so long as the other components do not change).

This is where you must change your thinking.

Flow rate is constant, pressure is not.

Correct.

[alacheesu]

Remember, pressure is a cumulative beast and each component will have a different PD depending on flow and where it is in the loop.

So pressure drop over a component is not just a function of flow, but also pressure at the inlet of the component?

[nikhsub1]

So pressure drop over a component is not just a function of flow, but also pressure at the inlet of the component?

Im not sure I understand your question. What I was trying to say is that pressure drop of a component is dependent on the pressure of the pump. If you use pump A with an EK Supreme it will have x pressure drop. Now, if you use pump B (which we will assume has 2x the pressure) the EK Supreme will exert MORE pressure loss with pump B and with pump A. How much more? Roughly 33% more (if the pressure is doubled). Pressure drop of a component is dependent on the pressure it is receiving and the resistance of the component itself. Pressure, flow and PD are VERY complex things to properly understand as there is no fixed target, values can change depending on many factors. What Vapor was saying is that no matter the position in the loop, each component will receive the same pressure and that is incorrect.

Edit* A quick way to demonstrate how PD increases with pressure; if you have a can of air (dust off, etc) take the red straw and blow though it gently. Now, blow as hard as you can. Notice the harder you blow the more resistance the straw makes? Same goes for water.

[Vapor]

What Vapor was saying is that no matter the position in the loop, each component will receive the same pressure and that is incorrect.

I never said that. I said that no matter where it is in a loop, it will have the same pressure drop--not that it will 'receive' the same pressure. I fully agree that pressure is not constant throughout a loop, each component (even tubing and fittings) has a given pressure drop at any flow rate.

If you're saying that position in a loop is a deterministic factor in pressure drop, then all the work that Swiftech, Thermochill, HWLabs, Martin, Koolance, Skinnee, et al have done with plotting pressure drop curves of their components is for naught.

But their work wasn't for nothing...at any given flowrate, a component has a certain pressure drop. There is no wiggle room with this. With that knowledge, a given set of components will have a combined net pressure drop (which is a curve, I agree that my 75-20-5 example was simplistic, but it's a lot easier to free-hand than using the actual curves) and with the PQ curve of a pump, you know the flowrate of the loop. Using the knowledge of the flowrate of the loop (which is constant throughout the loop), you know the specific pressure drop of every component and those values, again, do not vary depending on where they are in a loop.

EDIT: and pressure drop isn't a percentage function. It's a unit function.

[nikhsub1]

I never said that. I said that no matter where it is in a loop, it will have the same pressure drop--not that it will 'receive' the same pressure. I fully agree that pressure is not constant throughout a loop, each component (even tubing and fittings) has a given pressure drop at any flow rate.

If you're saying that position in a loop is a deterministic factor in pressure drop, then all the work that Swiftech, Thermochill, HWLabs, Martin, Koolance, Skinnee, et al have done with plotting pressure drop curves of their components is for naught.

But their work wasn't for nothing...at any given flowrate, a component has a certain pressure drop. There is no wiggle room with this. With that knowledge, a given set of components will have a combined net pressure drop (which is a curve, I agree that my 75-20-5 example was simplistic, but it's a lot easier to free-hand than using the actual curves) and with the PQ curve of a pump, you know the flowrate of the loop. Using the knowledge of the flowrate of the loop (which is constant throughout the loop), you know the specific pressure drop of every component and those values, again, do not vary depending on where they are in a loop.

EDIT: and pressure drop isn't a percentage function. It's a unit function.

OK, I misunderstood you then. You are correct that with the same components in the loop the LOOP PD will the the same no matter where they are placed. I thought you meant pressure to each individual component would be the same no matter where it is placed.

You said this:

Of course it does

But the pressure drop of a component is constant no matter where it is in a loop (so long as the other components do not change).

The above is what I was ranting about as this is misleading. We can agree that a CPU block right after the pump will get more pressure than right before the pump?

[Jah]

nah eric and dave... sadius is correct when it comes with the KL-350.

trust me ive had too much time on the bed with it.

Pump right b4 kl-350 on this block guys.

I'm interpreting this as putting the pump just before the block yield the best temperatures, and I have heard the same statement before.

Anyone have a theory to why?
As previously discussed the flow should be the same no mater what order we place the components and for all practical purpose water is incompressible, so the pressure on the inlet of the block should not really matter.
The only theory I can come up with is that you have a turbulent flow after the pump and if you place the block close enough you get turbulence at the inlet of the block for "free".

Anyone whit a better idea?

[Mech0z]

When disccusing this, would this be bad? RAD -> pump/res -> cpu -> gpu -> RAD it seems like the cleanest way of doing it.

[alacheesu]

Im not sure I understand your question.

I just found the statement that pressure drop depends on loop order a bit strange. I always thought pressure drop was a function of flow rate only. I think we all agree that the pressure at the inlet of the component depends on loop order, but the pressure drop (pressure at inlet minus outlet) over a component should be constant at a given flow rate, regardless of loop order.

[NaeKuh]

How much difference does it make? Wish I saw that before filling it and bleeding it LOL I suppose I will just leave it as is now unless the difference is really significant?

i recorded around a 2-3C benifit.

honda i think also reported around a 2C benifit.

I'm interpreting this as putting the pump just before the block yield the best temperatures, and I have heard the same statement before.

Anyone have a theory to why?
As previously discussed the flow should be the same no mater what order we place the components and for all practical purpose water is incompressible, so the pressure on the inlet of the block should not really matter.
The only theory I can come up with is that you have a turbulent flow after the pump and if you place the block close enough you get turbulence at the inlet of the block for "free".

Anyone whit a better idea?

realized lately my fluid mechanics is botched..

but originally, i anticipated the increase head pressure gave better turbulance inside the block, which then increased it efficiency.

Pressure at the inlet defines how fast and how effective your accelerators are.

[Snyxxx]

Pressure drop is dependent on velocity to the second power or V^2. Pressure drop increases much quicker with higher velocity.

Putting the block after the pump will result in more flow rate to the rest of the blocks downstream.

[Vapor]

The above is what I was ranting about as this is misleading. We can agree that a CPU block right after the pump will get more pressure than right before the pump?

Yeah, I agree it has more pressure at the head if it's the first thing in line. But because the drop is always the same, that also means more pressure leaves it....

My overwhelming feeling on this is that it means diddly for thermal performance of a block, but if tests indicate otherwise then that's what it is. I'll see if I can test this with my triple-DDC testbed sooner rather than later

[Martinm210]

I agree with Vapor on this one:

Pressure

Pressure does vary through the loop, it is at it's highest behind the pump and lowest right before the pump.

Pressure however has no impact to thermal performance, because liquids are incompressible at least in the sense of our tiny little pressures we're dealing with. If the fluid is not compressed, there no density change to the fluid, and it results in no performance difference.

Pressure Drop. Remember pressure drop is a "DIFFERENCE" in pressure, it is the pressure difference of the inlet vs. the outlet and it doesn't matter where in the system that component is. If the pressure drop was 5PSI at 1GPM, it would always be 5PSI. It could have 20PSI at the inlet, but it it was flowing 1GPM, the outlet would be 15PSI. Or it could be 7PSI at the inlet and 2PSI at the outlet.

Flow Rate
Flow rate is constant throughout the loop.

Velocity
Velocity is NOT constant throughout the loop, and to no suprise...velocity is the key to increased thermal performance. Because flow rate has to be constant throughout the loop, when an opening gets smaller, the fluid has to go faster. Nozzles and small channels are designed for one thing, to increase velocity where heat transfer occurs. The more velocity, the more turbulent the fluid and the more likely more of the fluid comes in contact with the heat transfering surfaces.

Loop Order
-The coldest water is located directly behind the radiator, but the differential is extremely small and almost always less than a degree and often only a few tenths of a degree cooler. The question you have to ask is, is this .2C worth the plumbing arrangement needed for that setup. For some the answer is yes, for many no. If the DTS sensor resolution is 1C, chances are you'd never be able to measure the difference.

-The only thing that really matters is having the reservoir or T-line before the pump so you can adequately fill the loop without the pump running dry at startup. After filling it really doesn't matter.

THIS WHOLE PUMP BEFORE THE BLOCK

HAS NEVER BEEN DOCUMENTED AS A BENEFIT. There's no scientific basis or reasoning that supports it. Sure blocks with high pressure drop often perform well, but it's not because of pressure. It's because of high velocities and the scrubbing that occurrs and that will be the same no matter where the pump is in the system

Don't believe me, you convinced it's "PRESSURE" that increases performance? Try this. You can add static pressure to a system by simply extending a T-line or fillport up the wall next to your PC and fill that tube completely up, you've just added static pressure to the whole system by that extra elevation. If pressure in itself makes a difference, this little trick by added pressure to the system should also improve performance.

But it simply won't because pressure by itself has no net benefit to thermal performance of water blocks. Velocity and forcing the water to impact those surfaces is really the only liquid property that provides benefit.

For most people loop order is insignificant enough not to worry about other than placing a res/t-line before the pump so they can fill the system. I still go with the common Rad>Block>Res>Pump scenario because I want that extra .2C, but I wouldn't bother if it screwed up tubing routing. I'd probably pay more attention though if you were running a TEC or a tri-SLI type of setup. If you had enough heat, you could start seeing more than a degree difference around the loop, but that would be pretty rare.

[prava]

Let me get this straight: so the best way to go for RL is still "do the shortest loop possible"? If I got what Martin said, it would mean that as Flow rate is constant throughout the whole loop, if you put a thinner tube you would have more velocity (but less...what? something has to decrease, nonsense otherwise) but nothing else and that's why a high restrictive block (normally jet-based) gets better performance, just because itself makes the water flow faster. Yai, this doesn't make much sense in my head at the moment, I admit. Pressure is force/surface, flow is liters (or gallons)/second yet what is speed in therms of water? The speed of each particle? I can see that with a different tubing you can change the velocity of water, yet I've got no idea how to quantify it....

[mrcape]

lol I'm surprised this debate is going on. It seems like something everyone setting up water cooling should ask. I'm finding that I have to figure it all out myself with water cooling because there are so many competing opinions in this sub forum.

Regarding pump before block I have had opposite suggestions and couldn't decide which way to go. I saw this thread and decided to go rad>cpu>pump and now I'm having second thoughts after overnight leak testing. There are no leaks, but the "clear" fluid looks like milk as it's getting churned up into foam water. I would think coldest water straight out of rad to cpu, but maybe not a good flow with the Koolance block?

I'm using the swiftech mcr320-res (which is kinda junky imo, bent inlets, fins etc.), Koolance CPU-350 and MCP355 with xspc top. Feser 1/2 ID 3/4 od tubing and feser fluid.

Questions -

Should I stick with this loop and let it bleed longer or reassemble with pump before cpu?

Should I bail on this swiftech res/rad and get a xspc 360 rad and a separate res?

Do they make a res that screws into the XSPC top or do I need a whole new "res top"?

Is it normal that radiator inlets are slightly bent and that they feel cheap overall?


I was really looking forward to this being quiet and cool and fun to work with and so far it's seemed like cheap toys. Maybe I need better parts? The guy from FrozenCPU advised me without checking and I ended up with a bunch of fittings that don't fit on my blocks - bitspower/too close together. I guess there is a bunch of trial and error with this stuff. In the future is there a place I can call that knows their stuff? This stuff is not rocket science for god's sake.

[alacheesu]

If I got what Martin said, it would mean that as Flow rate is constant throughout the whole loop, if you put a thinner tube you would have more velocity (but less...what? something has to decrease, nonsense otherwise)

Cross-sectional area decreases. Flow rate equals cross-sectional area times velocity, or Q=A*V. Flow rate, Q, must be constant everywhere, so by decreasing A (thinner tubing, nozzle in water block etc.), V must increase.

Yai, this doesn't make much sense in my head at the moment, I admit. Pressure is force/surface, flow is liters (or gallons)/second yet what is speed in therms of water? The speed of each particle? I can see that with a different tubing you can change the velocity of water, yet I've got no idea how to quantify it....

That's a little tricky. Individual water molecules will move in an unpredictable manner, so looking at just one won't work. What you do (one way to do it, at least) is imagine a tiny cube placed somewhere in the fluid flow and look at the average net number of water molecules entering and exiting each face of the cube. Skipping the rest of the details, this gives you a way to say in which direction, and how quickly, the water in the cube is moving on average. There's a lot more to this, but that would make a very long post not very relevant to water cooling. :P

[prava]

Cross-sectional area decreases. Flow rate equals cross-sectional area times velocity, or Q=A*V. Flow rate, Q, must be constant everywhere, so by decreasing A (thinner tubing, nozzle in water block etc.), V must increase.


That's a little tricky. Individual water molecules will move in an unpredictable manner, so looking at just one won't work. What you do (one way to do it, at least) is imagine a tiny cube placed somewhere in the fluid flow and look at the average net number of water molecules entering and exiting each face of the cube. Skipping the rest of the details, this gives you a way to say in which direction, and how quickly, the water in the cube is moving on average. There's a lot more to this, but that would make a very long post not very relevant to water cooling. :P

THX!!!! Ahhh ok. So I already said that, just didnt realise what it was, hehe, what decreases is the area in which the water is moving (perpendicularly, of course) across...yet simple and clear

About speed...yes, I remember that when I was in high-school (physics class) when we used that stuff to calculate current as you use the same when using fluids...mmm can't remember the name of that "principle" or "law", my memory is not what used to be.

BTW, there is something going around, why do we need any pressure at all then? If the only thing we need is more flow (we can modify the speed of the water at our wish just modifying the width of the tube), when does pressure get into the equation? As far as I know, pressure would be useful in a non-closed loop when you've got a well and you need to pump up the water, but in this case I don't see the need for it...I know, i'm missing something again.

PD: yai, although not being that technical, having to understand something not that simple in a language you don't use in your everyday living makes everything more complicated (plus the inch/cm - liter/gallon stuff).

[Martinm210]

Let me get this straight: so the best way to go for RL is still "do the shortest loop possible"? If I got what Martin said, it would mean that as Flow rate is constant throughout the whole loop, if you put a thinner tube you would have more velocity (but less...what? something has to decrease, nonsense otherwise) but nothing else and that's why a high restrictive block (normally jet-based) gets better performance, just because itself makes the water flow faster. Yai, this doesn't make much sense in my head at the moment, I admit. Pressure is force/surface, flow is liters (or gallons)/second yet what is speed in therms of water? The speed of each particle? I can see that with a different tubing you can change the velocity of water, yet I've got no idea how to quantify it....

No, don't worry about the loop speed, the ONLY place where high velocity is a benefit is where heat exchange occurs. The tubing and everything else in the loop doesn't matter, in fact you get less resistance the larger the openings and the slower the flow is everywhere else. This is why it's beneficial to run larger tubing, etc...Generally though, there's very little you can do regarding the loop to make any significant gains, it's even pretty hard to make one bad really. I would just recommend having fun with it and lower restricting in the tubing and fittings is always a good thing, but it's not worth going completely out of your way for.

Block design is a careful balance of factors, restriction is just one of those. You don't want a block so restrictive that flow rates are too low, but you also don't want something so open the water has a chance to bypass without transferring heat. But there are many other factors as well like surface area, collecting and distributing that heat, etc. Generally don't worry about it, that's what the block makers do for us.

Yes, I was just noting flow rate is constant throughout the loop, so velocity is a factor of the opening sizes. You'd like big opening for low restriction where it doesn't matter and save that energy for the blocks and radiators where it's helpfull. Smaller area means the water has to move faster to maintain that constant flow rate is all. Velocity increases is done with nozzles and fine copper channels, etc.

lol I'm surprised this debate is going on. It seems like something everyone setting up water cooling should ask. I'm finding that I have to figure it all out myself with water cooling because there are so many competing opinions in this sub forum.

Regarding pump before block I have had opposite suggestions and couldn't decide which way to go. I saw this thread and decided to go rad>cpu>pump and now I'm having second thoughts after overnight leak testing. There are no leaks, but the "clear" fluid looks like milk as it's getting churned up into foam water. I would think coldest water straight out of rad to cpu, but maybe not a good flow with the Koolance block?

I'm using the swiftech mcr320-res (which is kinda junky imo, bent inlets, fins etc.), Koolance CPU-350 and MCP355 with xspc top. Feser 1/2 ID 3/4 od tubing and feser fluid.

Questions -

Should I stick with this loop and let it bleed longer or reassemble with pump before cpu?

Should I bail on this swiftech res/rad and get a xspc 360 rad and a separate res?

Do they make a res that screws into the XSPC top or do I need a whole new "res top"?

Is it normal that radiator inlets are slightly bent and that they feel cheap overall?


I was really looking forward to this being quiet and cool and fun to work with and so far it's seemed like cheap toys. The guy from FrozenCPU advised me without checking and I ended up with a bunch of fittings that don't fit on my blocks - bitspower/too close together. I guess there is a bunch of trial and error with this stuff. In the future is there a place I can call that knows their stuff? This stuff is not rocket science for god's sake.

You'll find that most of the highly debated topics are generally those that are either hard to measure, nobody has tested them, or they are generally insignificant. I didn't believe it much myself until I install a thermal probe before and after my CPU block. I got bored looking at the same insignificant difference. Temp out was always just a couple of tenths of a degree higher than in...not very interesting.

The bubbles could either be a need for time (High restriction loops can take several days to completely bleed out. Or it could be a reservoir that's sucking air back into the loop. Make sure your QPres rad is as full as possible and vertically installed as required. Also rare, but it's possible to suck air at the inlet of a pump because this is the point where the least pressure is and it's possible to even have negative pressure if the reservoir is too far away and the pump is place high up in the system.

Regarding the QC issues, mass production for you there, that's what gives you the low price. If it's bad enough though, contact swiftech, they've been good about RMA's from what I can tell. You generally get a little higher level of QC on the more expensive rads, but you sure pay for it.

I'd recommend calling Petra's or Sidewinder next time, they've always been really helpful for me.

BTW, there is something going around, why do we need any pressure at all then? If the only thing we need is more flow (we can modify the speed of the water at our wish just modifying the width of the tube), when does pressure get into the equation? As far as I know, pressure would be useful in a non-closed loop when you've got a well and you need to pump up the water, but in this case I don't see the need for it...I know, i'm missing something again.

PD: yai, although not being that technical, having to understand something not that simple in a language you don't use in your everyday living makes everything more complicated (plus the inch/cm - liter/gallon stuff).

We only need some pressure to keep our tubes formed, otherwise what we really need is "Pressure Difference". That's what pumps give us, a pressure difference from the outlet back to the inlet, and it's this pressure difference that's the form of energy that moves the fluid around. That's all it is, the pump produces the pressure differential, then the blocks and rest of the system absorb that pressure energy as the fluid moves around the system. Pressure difference is the only thing (Besides gravity) that can move a fluid.

Even in an open loop system, the pump is still just producing a pressure difference, but there is a distinct difference. In an open loop you can gain or loose static pressure energy. If you pump water up 100' into a tank, it now has alot of stored energy in the form of elevation. In a closed loop, any water that is pumped up, comes back down so static pressure gains and losses net 0. All we're left with is pressure drop (difference across blocks, etc due to friction), and pressure gain (difference across the pump). Both of these have a performance curve. We just refer to the blocks/fittings as pressure drop. And it's also more common to "Think" of pump as pressure head. But in the end what's really going on is pressure differences, the actual pressure numbers themselves are not important, only the difference.

Not sure that helps at all, but it's all only as complicated as you want it to be.

For most systems, deciding on a pump can be as simple as picking a good strong one like a D5 or DDC with top. It's pretty unlikely that you'll pick components that are incompatible with either of these.

Then you have your tubing to pick from. Bigger is better, but pretty insignificant, go with what you like and don't worry about it.

Then there's the blocks. They are all very good anymore. Of the 15 block I tested on the Q6600, they were all within 4 degrees of each other and they ALL cooled my processor to the OC that I wanted.

Now the radiators. I also don't think you can pick a bad one. The only thing here is having enough. My recommendation, GO BIG, this will let you run lower speed fans. EI, don't go buy a single 120 sized radiator and expect you can run 1000RPM fans on it and cool several hundred watts. The fan and radiator are a duo package it's the combo that dissipates heat and they are both highly dependant on each other.

Loop order: Pretty simple, put a reservoir or T-line before the pump so you can fill it. If you care, it's a little better to have the radiator before the block so you have the very coldest water hitting the block. Otherwise, it's pretty flexible.

So...in the end. It's pretty hard to screw up a water cooling loop. Most of the good brand name stuff is really good. About the only real area that seems to have significance for most loops I've seen is regarding the radiator and fans. If the average water temperature is about 10 degrees hotter than the air going into the radiator, you've got about 10 degrees you can work with and reduce by either installing more radiator or stronger fans.

The rest is pretty well defined and not much you can do with it.

[mrcape]

Thx for the advice Martin. I'm going to give this loop a little more time and see if it clears up. The Res/rad is vertical and seems to not be leaking at all. Since this is supposed to be a restrictive cpu block I think I'll get a whole separate res / pump / rad / block for my NB. At least then I can really compare the performance of the stuff and learn about the current products.

[DeathWalking]

That's a little tricky. Individual water molecules will move in an unpredictable manner, so looking at just one won't work. What you do (one way to do it, at least) is imagine a tiny cube placed somewhere in the fluid flow and look at the average net number of water molecules entering and exiting each face of the cube. Skipping the rest of the details, this gives you a way to say in which direction, and how quickly, the water in the cube is moving on average. There's a lot more to this, but that would make a very long post not very relevant to water cooling. :P

Oh my god, please tell me Gauss' Law does not apply to fluid dynamics!

[prava]

Oh my god, please tell me Gauss' Law does not apply to fluid dynamics!

Gauss, it was Gauss!! If I don't remember it wrong this is what our physics teacher told us, that the same rules apply for liquids and current, and so we use the same law (Gauss' one).

We only need some pressure to keep our tubes formed, otherwise what we really need is "Pressure Difference". That's what pumps give us, a pressure difference from the outlet back to the inlet, and it's this pressure difference that's the form of energy that moves the fluid around. That's all it is, the pump produces the pressure differential, then the blocks and rest of the system absorb that pressure energy as the fluid moves around the system. Pressure difference is the only thing (Besides gravity) that can move a fluid.

Even in an open loop system, the pump is still just producing a pressure difference, but there is a distinct difference. In an open loop you can gain or loose static pressure energy. If you pump water up 100' into a tank, it now has alot of stored energy in the form of elevation. In a closed loop, any water that is pumped up, comes back down so static pressure gains and losses net 0. All we're left with is pressure drop (difference across blocks, etc due to friction), and pressure gain (difference across the pump). Both of these have a performance curve. We just refer to the blocks/fittings as pressure drop. And it's also more common to "Think" of pump as pressure head. But in the end what's really going on is pressure differences, the actual pressure numbers themselves are not important, only the difference.

Thx Martin for your time, I still remember the very first time I came across your reviews and i was like "No way, those graphs have to mean something but yet I've got no clue about what the hell they do mean". Now everything seems a bit more simpler...thanks to you and other people who invested their precious time trying to show the others how do thinks really work.

Ok. So pressure is used in order to make the water "move" around the loop. If we get it to 0, then there is no flow, in some ways it's like friction because it opposes to the movement...yet it depends on it to be there (as there is no friction is there is nothing trying to move).

OK, to sum up and to check if i've got everything alright (if I haven't I'm gonna commit suicide, i swear
So, when you choose a pump, the only think that really matters is the flow rate...and the pressure if you would exceed it, otherwise it just doesn't matter at all. The shortest the circuit the better because you reduce pressure drop (as everything in the circuit reduces pressure). And, the other doesn't matter at all cause the flow is constant throughout the loop (BTW, you said that rad after pump is a good idea because doing so the cpu block will get the very coldest water... but now that I think about it, the difference might be totally negligible as the pump would only put about 20W in the water, which is nothing compared to all the other heat other blocks put in...).
Mmmm which means that a D5 is the way to go if you aren't going to put many blocks or they aren't really restrictive and the DDC is better if you are putting plenty of blocks or a hell of a restrictive one. BTW, is there any way to mount a loop (a possible one, not one with 10 blocks) with a D5 in a way that the total pressure drop exceeds the head pressure and so water just doesn't flow? Because if there isn't (or at least not one that you will ever do) I don't see the point about not buying the D5...

You'll find that most of the highly debated topics are generally those that are either hard to measure, nobody has tested them, or they are generally insignificant. I didn't believe it much myself until I install a thermal probe before and after my CPU block. I got bored looking at the same insignificant difference. Temp out was always just a couple of tenths of a degree higher than in...not very interesting.

But that's something that doesn't need rocket science to understand (if you compare it to all the other stuff) If you just take in account how much water flows at that point it's pretty clear that you will need a hell of a good instrument to measure the difference between inlet-outlet temperature.
I guess it's the only part I could understand precisely (not much, but hey, its a start).

[Erasmus354]

OK, to sum up and to check if i've got everything alright (if I haven't I'm gonna commit suicide, i swear
So, when you choose a pump, the only think that really matters is the flow rate...and the pressure if you would exceed it, otherwise it just doesn't matter at all. The shortest the circuit the better because you reduce pressure drop (as everything in the circuit reduces pressure). And, the other doesn't matter at all cause the flow is constant throughout the loop (BTW, you said that rad after pump is a good idea because doing so the cpu block will get the very coldest water... but now that I think about it, the difference might be totally negligible as the pump would only put about 20W in the water, which is nothing compared to all the other heat other blocks put in...).
Mmmm which means that a D5 is the way to go if you aren't going to put many blocks or they aren't really restrictive and the DDC is better if you are putting plenty of blocks or a hell of a restrictive one. BTW, is there any way to mount a loop (a possible one, not one with 10 blocks) with a D5 in a way that the total pressure drop exceeds the head pressure and so water just doesn't flow? Because if there isn't (or at least not one that you will ever do) I don't see the point about not buying the D5...

No flow rate is not the only thing that matters when choosing a pump. Head pressure is still very important. If you look at pump PQ charts you will notice it is a curve. The pump produces different flow rates at different pressure differentials. In general pumps with higher maximum head pressure will get more flow in a water cooling loop when you factor in the pressure differential.

If you were to graph the pressure vs. flow of all your watercooling components combined and superimpose that on the pump pressure vs. flow chart where the two lines intersect is the flow rate you will get in your loop. The Watercooling estimator spreadsheets show this relationship very well. So you want to chose the pump that has the best flow rate for your components (intersects the pressure vs. flow curve of your components at the highest flow rate).


As for temperature drop from a radiator. A 400Watt loop with 4lpm flow (~1gpm) will have a temperature drop of 1.43 degrees celsius from input to output. And you are correct in pointing out how little heat the pump dumps into the water. In that same loop the 20W heat dump pump would up the water temperature by only ~0.07 C. If you assume 200W cpu and 180W gpu and 20W pump if you were to put the radiator directly AFTER the cpu you are losing ~0.7C in water temperature as opposed to putting the radiator directly before the CPU.



@Martin: Great explanation, you put it much more succinctly than I ever could have.

@Naekuh: o.O never thought I would see the day.

[Jokester_wild]

res>pump>rad>pump>cpu>gpu /end

[SNiiPE_DoGG]

I thank martin for his time in sorting this all out


the rest of this thread is just pointless - people throwing around their unfounded opinions on the matter....