One reservoir for three loops

[sewi]

Hi,

One reservoir for three loops? Is that going to work right? My idea is to save space needed for reservoirs and also to equalize the water temp between loops so that when GPU is not used the second loop will actually help to cool down the processor.
Are there any cons?
Or maybe I could run 2 loops for blocks and the third one for rads. Just an idea.

[Church]

Do as NaeKuh did. For least important loop with MB he put some minires with integrated pump. It's compact enough and leaves wide options of selecting some res for dual loops, eg. XSPC ones. Though do you really really need 3 loops?

[Sadasius]

Then it is not 3 loops. If it uses one res then it's one big loop. Better off making it better as one big loop and take out a pump and serialize the 2 left over pumps and make the one loop a good one. Cut down on unnecessary tubing by making the shortest runs and it will look nice and clean ta boot!

[sewi]

Its likely i will never really use three loops and three rads. But its xtreme systems right?
It will be like one loop but there will still be only like two block and one rad per pump. Which will give some good flow rate. Guess only.

[Waterlogged]

Its likely i will never really use three loops and three rads. But its xtreme systems right?
It will be like one loop but there will still be only like two block and one rad per pump. Which will give some good flow rate. Guess only.

No, like Sadasius already said, it will be one loop. Anytime there is a common component between two "supposedly separate" loops, it become one loop. In this case, and every other case for that matter, the common component is the res.

[sewi]

Ok, I don't wanna argue about that. The point is that one pump will only have two blocks and one rad. Plus if one stream will get hotter because of higher power income the water will mix up so that the pump will always take avarage cool wter from res and pump it again trough blocks.

[Alexandr0s]

Ok, I don't wanna argue about that. The point is that one pump will only have two blocks and one rad. Plus if one stream will get hotter because of higher power income the water will mix up so that the pump will always take avarage cool wter from res and pump it again trough blocks.

Then save yourself the trouble and place everything in a single loop, it'll work the same.

[Church]

Only gain on having such pseudo separated loops = less restriction for pumps to work with. You loose other gains of separate loops like temperature separation/isolation and such, yet still having minuses like more difficult/cluttered tubing and duplicated components. I'd say go classic separate loops or single loop. Middle ground will bring more problems then gains with rare exceptions (like typhoon III D5 res/top which somehow managed in those pseido loops bring up total flow sum way more then one D5 pump should (i'll leave out quality issues with those res-es though :) )).

[PiLsY]

If you can fit it all in then go for it. I see what you're trying to do and in my opinion you'll get slightly better temps than a single series loop.

Main factor as always is cool air to the rads.

[Cue]

No, like Sadasius already said, it will be one loop. Anytime there is a common component between two "supposedly separate" loops, it become one loop. In this case, and every other case for that matter, the common component is the res.
don't shoot me, but can I ask why? I genuinely would like to know the scientific reason behind this, I remember a very good diagram of another way of thinking about the connected loops which showed the reason but I can't seem to find it anymore. I'm a physicist so don't hold back with the explanation, though probably not a very good one :D.

[NaeKuh]

I'm a physicist so don't hold back with the explanation, though probably not a very good one :D.

Multi loops are independant of each loop, so they have there own holding capacity.

While a connected loop will share holding capacity.

Meaning when your gpu goes up on a single loop, the res temp will rise as well as your overall coolant level.

However in a real multi loop, when your gpu gets loaded, your cpu is unaffected because its on a completely different system.

When you boil down to multi and single loops, its about control.
How much control do you want in each independant loop.

The way you have it, its just best to go serial, or put each pathway as paralell fed by a pump.


Do as NaeKuh did. For least important loop with MB he put some minires with integrated pump. It's compact enough and leaves wide options of selecting some res for dual loops, eg. XSPC ones. Though do you really really need 3 loops?

You mean this guy?

http://i125.photobucket.com/albums/p73/aigomorla/Haruhi/IMG_1378.jpg

its a PA160.. and if anyone tells u a PA160 is not enough for a board loop, they have to be smoking some really good stuff.. or they have an ASUS Super Nova Series... meaning something which no one has which heats up as much as the sun. :P

[Holst]

There may be some flowrate benefits from running your loop like this.
Combined res and 3 rads should ensure best use of the rads and low water temps (as long as you get the rad placement right)

However its hard to calculate what the best configuration will be.

The best bet will be to look at skinnee's graphs for your blocks and work out if there will be any significant gains putting them on there own independent mini loop or not.
If you have one block that is comparatively high restriction it may be better to run that on its own pump to maximize flow through your other components.

It may be that the best layout for cooling becomes a tubing nightmare.

[Manicdan]

can you break down what your trying to cool and what parts you have (cpu block, # of gpu blocks, number and sizes of radiators, etc)

i do agree that it would be nice to have more cooling power for the hot component with the other ones are idling, AND it helps with proper load balancing. like if you have a 120x3 for a cpu that wont break 200W, then 2x 580s which might have their own 120x3, obviously in gaming the gpus will be getting quite warm and the cpu will be laughably cold.

the argument of one loop vs 3 is just about pump pressure per block, rather than temperature sharing i guess. so while the radiators are treating it like one big loop (water shared at the res) the pumps can handle it like 2 or 3 loops to keep flow going.

and about any kind of tubing issues, i think it can look really cool and messy rather than clean and open.

[NaeKuh]

and about any kind of tubing issues, i think it can look really cool and messy rather than clean and open.

tubing issues??? WHERE?

Tri Loop:
http://i125.photobucket.com/albums/p73/aigomorla/Haruhi/IMG_1687.jpg

http://i125.photobucket.com/albums/p73/aigomorla/Haruhi/IMG_1385.jpg


Planning is what it takes... that is 3 loops.
1 -> CPU
2 -> GPU
3 -> Board..

Independent loops on 4 radiators:
http://i125.photobucket.com/albums/p73/aigomorla/Haruhi/IMG_1386.jpg

i dont exactly call that a spiders web... however if it was all connected via one loop, i can say it would of been.

[Alexandr0s]

Just curious: Why is there a piece of open tubing hanging in the case? :p:

[Manicdan]

Just curious: Why is there a piece of open tubing hanging in the case? :p:

it the overflow port just like a car radiator has

water expands when hot, so he has it dump out any excess water to keep the radiator from bursting when his temps reach near/over 100C

duh....

[Waterlogged]

If you have a loop with 2 pumps/2 rads or 20 pumps/20 rads, anytime the water has a chance to mix with the "other loops", it's one single loop. There must be complete segregation of all components and water for the loop to be counted as a individual loop.

[NaeKuh]

Just curious: Why is there a piece of open tubing hanging in the case? :p:

excess fill, so i make sure all the loops are completely bleed.

Usually my builds now all have a bit excess off the res, or leading to the res for fill.

http://i125.photobucket.com/albums/p73/aigomorla/IMG_0005.jpg

Its so my res's dont do the crazy cyclone action... but meh.. wont do that anyways cuz theres a bar of silver inside that res which that tube connects to like above.

[skinnee]

sweet... I needed some Naekuh photo's to turn my day around.

[Captain_Harlock]

My God Naekuh, is that rust on your castor bolts?

[Church]

NaeKuh: you have no tubing issues, what you have - it's dust issues. :wasntme: :sofa:

[KingOfsorroW]

If you have a loop with 2 pumps/2 rads or 20 pumps/20 rads, anytime the water has a chance to mix with the "other loops", it's one single loop. There must be complete segregation of all components and water for the loop to be counted as a individual loop.

Thanks for the clear straight forward answer. I was also wondering what would happend if one reservoir would be connected to several loops. It is clear now it shouldn't be done. If you want to keep the loops separated, of course.

[Manicdan]

Thanks for the clear straight forward answer. I was also wondering what would happend if one reservoir would be connected to several loops. It is clear now it shouldn't be done. If you want to keep the loops separated, of course.

there are still different reasons for keeping them together or separate and i think some standardized terminology should be established

what should it be called when you have 2 loops that share a reservoir vs 2 loops that are independent.

keeping one res does let your temps blend, but it does keep your pump pressure separate, which might be the only reason for separating loops.

[Captain_Harlock]

parallel loop(s)

[KingOfsorroW]

there are still different reasons for keeping them together or separate and i think some standardized terminology should be established

what should it be called when you have 2 loops that share a reservoir vs 2 loops that are independent.

keeping one res does let your temps blend, but it does keep your pump pressure separate, which might be the only reason for separating loops.
So what's the point in blending temps then? Temps is the main thing here - the main reason why we watercool. By blending you maen inccreaing the minimum temps and, probably, the decrease of the max temps, but not necessarily - they may increase as well, I suppose. So, who would need such a blend - only someone with very low temps in all loops, but then the question arise - whats the point in keeping the loops separated at all? :shrug:

[Tetrafluorometh]

Tom over at Chilled PC did 3 loops in a Lian Li PC70. All with Thermochill PA's! T lines, small, quick, simple!

http://www.chilledpc.co.uk/forum/showthread.php?t=2443

[Manicdan]

So what's the point in blending temps then? Temps is the main thing here - the main reason why we watercool. By blending you maen inccreaing the minimum temps and, probably, the decrease of the max temps, but not necessarily - they may increase as well, I suppose. So, who would need such a blend - only someone with very low temps in all loops, but then the question arise - whats the point in keeping the loops separated at all? :shrug:

the first post has a very simple point


My idea is to save space needed for reservoirs and also to equalize the water temp between loops so that when GPU is not used the second loop will actually help to cool down the processor.

if you have 2 smaller rads that can do an OK job cooling down their own components, if you put them together you can use it as a load balancer, without sacrificing pump pressure.

the only drawback is that the more temperature sensitive component is going to be your limit. so the trick to solve that is base your fans speeds on that component. if your cpu goes above 50c increase fan speeds, if your gpu goes above 60c, you dont care cause your cpu my not be loaded enough to go beyond its stable threshold.

this kind of idea works well when you know how many programs load both your cpu and gpu heavily. playing games will use about 70% of the gpu (depending on brand and game) and maybe only 30% of the cpu. but if you do encoding in the background you would run into issues.

this idea also works well if both components use similar TDPs, or if the one with the weakest power draw also is not temperature sensitive. like the VRMs on the MB and its NB might only add 40w to the mix, and dont care if they get hotter water. but if you have 1000w in gpu power, and a 100w cpu, then the cpu will always be dumped hot water weather or not its being loaded, and the gpus wouldnt even notice a temp change if the cpu is loaded or not.

there are definitely ALOT of pros and cons to weigh, but in the end it can be beneficial

[Biffa]

I've got a hybrid loop. Two pumps, one pushing to the cpu, mb block then back to the res, the other pump pulling from the same res to the gpus and back. Works for me :)

[Cue]

the first post has a very simple point
if you have 2 smaller rads that can do an OK job cooling down their own components, if you put them together you can use it as a load balancer, without sacrificing pump pressure.

the only drawback is that the more temperature sensitive component is going to be your limit. so the trick to solve that is base your fans speeds on that component. if your cpu goes above 50c increase fan speeds, if your gpu goes above 60c, you dont care cause your cpu my not be loaded enough to go beyond its stable threshold.

this kind of idea works well when you know how many programs load both your cpu and gpu heavily. playing games will use about 70% of the gpu (depending on brand and game) and maybe only 30% of the cpu. but if you do encoding in the background you would run into issues.

this idea also works well if both components use similar TDPs, or if the one with the weakest power draw also is not temperature sensitive. like the VRMs on the MB and its NB might only add 40w to the mix, and dont care if they get hotter water. but if you have 1000w in gpu power, and a 100w cpu, then the cpu will always be dumped hot water weather or not its being loaded, and the gpus wouldnt even notice a temp change if the cpu is loaded or not.

there are definitely ALOT of pros and cons to weigh, but in the end it can be beneficial

Thanks for the great explanation. This clears up all my suspicions. I had read gabes serial vs dual loop thread

http://www.xtremesystems.org/forums/showthread.php?t=254683&highlight=pumps+serial

Which sort of covers the same theory about load balancing. I think going the parallel loops route with a shared res is the same as going from dual loops to single loop when it comes to load balancing so you can compare it to gabes results. but it adds the advantage of greater flow with restrictive loops which can mean better heat transfer in some blocks, so even better temps, as long as the difference in tubing length between serial or parallel loops is not too high that is.

IMO parallel loops is the best route to go for most use cases. With a temp controller it can also mean less noise when only one component is under stress or when completely idle.

As for my current components, I currently have 2 individual loops

MCR Res>D5 Vario>GTX120>EK Supreme>GTX120>inline Flow Meter>inline Temp Sensor

MCR Res>D5 Vario>EK VGA Full Cover block>GTX360>inline Flow Meter>inline Temp Sensor

unfortunately one of my MCRs is cracking really badly so I've ordered a Koolance 452X2. Since the 452X2 has so many options it made me rethink about my loop(s). I've almost settled for parallel but I have 8/10mm tubing so serial is also an option I've been thinking about to cut down on the amount of head loss.


serial
452X2 Res> D5 Vario x2> GTX120> EK Supreme> EK VGA GTX120+GTX360>inline Flow Meter>inline Temp Sensor

Advantages
less components and less tubing
only 3 quick disconnects needed
Redundancy

Disadvantages
VGA block straight after CPU. does this matter given that water has a high specific heat capacity and the flow is sufficient? would it increase my VGA temp or is it negligible.

[Manicdan]

Disadvantages
VGA block straight after CPU. does this matter given that water has a high specific heat capacity and the flow is sufficient? would it increase my VGA temp or is it negligible.

someone whos done alot of testing should know that answer easily

i would just touch them with your hand and see if you can tell the difference between the inlet and outlet of the cpu block. or use a thermal probe of some kind for exacts.

higher flow does help this problem significantly

[NaeKuh]

im laughing right now..

Because ultimately the main point in multi loops is control.

How much control do you want in your system?

If your not a xtremeist like some of us are, then a single loop isnt bad.
Its better then air, but our xtremeist in this forum will rock you in temps.

Are you trying to mirror our xtremeists? or you trying to get work done with the smallest dent in your pocketbook?

[coolmiester]

Damn.........reading this thread my triple loop SR-2 build just got relegated to being called a dual loop :rofl:

Forgive me but i'm gunna keep calling it a triple loop.

[Waterlogged]

Lets agree on definition of a loop first, I think this should suffice. ;)

http://www.thefreedictionary.com/Loop

Now going by a few definitions there, a loop is is something that repeats itself...so, using that definition.

In the image below are several of the common type loops we've seen over the years. Component count will vary per loop but I've set it up in a manner to try and keep it a bit simple.

Loop Ex.1 is your basic single loop with everything included in series.

Loop Ex. 2 is your basic dual loop with the major heat producing parts each having their own loop. This pattern can be repeated for each component that has a block made for it as long as there is room for all the parts. We actually had 1 guy try to fit 7 loops into a case at one point.

Loop Ex.3 is technically a single loop even though each loop has a complete set of parts. This is because the water itself is allowed to flow through either loop and is the factor that links them into a single loop. This setup is commonly mistakenly called a parallel loop. There is also a variation on this with only 1 pump that drives everything, this too is a single loop.

Loop Ex. 4 is another variation on the "parallel" loop, again, it too it a single loop due to the water being able to flow through either path.

http://img20.imageshack.us/img20/5733/loopssw.jpg

There are a few reservoirs out there that do allow true dual loops such as the split DDC res from XSPC or the new res's from Koolance or the single bay Delrin res's that Primochill did get from BoxGods.

HTH clear up some of the cloudiness that surrounds this topic.

[coolmiester]

Loop Ex.3 is technically a single loop even though each loop has a complete set of parts. This is because the water itself is allowed to flow through either loop and is the factor that links them into a single loop. This setup is commonly mistakenly called a parallel loop. There is also a variation on this with only 1 pump that drives everything, this too is a single loop.


Yep, i'll take that explanation as it mentions each loop in the singular form at least twice indicating there is obviously more than one loop so my dual loop just got promoted back to triple loop :up:

[Waterlogged]

Yep, i'll take that explanation as it mentions each loop in the singular form at least twice indicating there is obviously more than one loop so my dual loop just got promoted back to triple loop :up:

Selective reading? :shrug:

[Cue]

Lets agree on definition of a loop first, I think this should suffice. ;)

http://www.thefreedictionary.com/Loop

Now going by a few definitions there, a loop is is something that repeats itself...so, using that definition.

In the image below are several of the common type loops we've seen over the years. Component count will vary per loop but I've set it up in a manner to try and keep it a bit simple.

Loop Ex.1 is your basic single loop with everything included in series.

Loop Ex. 2 is your basic dual loop with the major heat producing parts each having their own loop. This pattern can be repeated for each component that has a block made for it as long as there is room for all the parts. We actually had 1 guy try to fit 7 loops into a case at one point.

Loop Ex.3 is technically a single loop even though each loop has a complete set of parts. This is because the water itself is allowed to flow through either loop and is the factor that links them into a single loop. This setup is commonly mistakenly called a parallel loop. There is also a variation on this with only 1 pump that drives everything, this too is a single loop.

Loop Ex. 4 is another variation on the "parallel" loop, again, it too it a single loop due to the water being able to flow through either path.

http://img20.imageshack.us/img20/5733/loopssw.jpg

There are a few reservoirs out there that do allow true dual loops such as the split DDC res from XSPC or the new res's from Koolance or the single bay Delrin res's that Primochill did get from BoxGods.

HTH clear up some of the cloudiness that surrounds this topic.

This is all semantics but why is loop 3 not to be referred to as a parallel loop? For example the same configuration would be called a parallel circuit if it were electrons instead of water, and by your definition of a loop in the link it has "electricity, closed circuit" as a definition. It would seem calling it a parallel loop is logical, if we shouldn't call it a parallel loop what do you suggest we call it?
This is a genuine question and not to be taken as a condescending/rhetorical question. I really want to know what to call it in future.

Also loop 4 is not at all a variation of 3; it's serial through and through unless those two res are referring to the same res, in which case it IS loop 3 just drawn differently.



If your not a xtremeist like some of us are, then a single loop isnt bad.
Its better then air, but our xtremeist in this forum will rock you in temps.

If you know some taliban members with chilled water loops in these forums then yeah I probably would get rocked (stoned?) in temps. :D
If however by xtremists you mean those with dual loops (which I currently have but contemplating going serial/parallel) I'd have to curiously ask, why? Because from the evidence in gabes tests it seems I would get better temps under most loads, with the amount of radiators I have, if I were to go serial/parallel.

Granted if I were pushing my volts and clocks to the point were my temps are 1.8 degrees away from my CPU exploding then yeah the separate loop would give me the control to push it up to that dangerous limit independent of my GPU, but seeing as I'm not then it would seem serial/parallel is the way to go. correct? Is this what you're referring to when you say control?


Are you trying to mirror our xtremeists? or you trying to get work done with the smallest dent in your pocketbook?

I'm not sure what part of buying a £125 ($200) res gave you that idea NaeKuh ;). I'm not worried about my wallet (pocketbook?). I'm after what most people are when they switch to water cooling

better temps, reliability/convenience, silence

I thought watercooling isn't all that extreme anyway. It's just more convenient and reliable. :shrug:

The sort of work I do on my system is number crunching (Finite element analysis to be specific) so I leave my PC on for extended periods, therefore

silence is a blessing when working
need good reliability/low error because the results are mission critical so overclocking too far is out of the question anyway.

do you suggest I go dual instead of parallel/serial then? Please give me an answer that is not overzealous to a specific method.
To be honest I could probably easily test both parallel and dual with the 425X2 that arrived today without having do much, but please provide a reason why serial is definitely out of the question if I want good temps.

I'll post some pics of the fantastic 452X2 res and maybe some test results as soon as I can upload them.

[Waterlogged]

This is all semantics but why is loop 3 not to be referred to as a parallel loop? For example the same configuration would be called a parallel circuit if it were electrons instead of water, and by your definition of a loop in the link it has "electricity, closed circuit" as a definition. It would seem calling it a parallel loop is logical, if we shouldn't call it a parallel loop what do you suggest we call it?
This is a genuine question and not to be taken as a condescending/rhetorical question. I really want to know what to call it in future.

Also loop 4 is not at all a variation of 3; it's serial through and through unless those two res are referring to the same res, in which case it IS loop 3 just drawn differently.

All the drawings are of a start to return to start variety so yes, the res is the same res. As for the "parallel", the runs are parallel but the loop itself isn't. Call it semantics if you wish but it's actually "technically" correct. . .can you "technically" turn off gravity? ;) I really don't know what to call the loop. :shrug:

Ex. 3 is a single feed out of the res to both pumps via a T or Y, the same res (therefore water) feeds both pumps which is the common link and negates the ability of calling them separate loops. Admittedly, this version is rather rare these days but does still show up from time to time.

Ex. 4 is separate feeds for each pump coming from the res and is exactly as you described it, "drawn differently", it is the same design minus the T's or Y's. I drew both versions because there are some res's out there that only have enough holes for 1 in/1out (Ex.3) and others have 2 in/2 out (Ex.4).

[Biffa]

I was well aware with my setup, albeit before the split XSPC bayres was available, that I was in reality doing a single loop, but the increase in flow rate by having individual pumps pushing to each section of the loop swung it for me.

So mine is Ex. 3 in your diagrams, although there is no Y connectors on the feed or return as the pumps are built into the XSPC res. (old version single res) plus I cross the flow (re Ghostbusters) on the return to the res. i.e. the CPU side returns to the GPU pumps feed and vice versa.

Seems to work ok but silent its not, but its a hot beastie doing hot work with not great ambients :)

[Martinm210]

Just to go against the grain, it's not a single loop...it's separate single pumping loops with a shared reservoir/heat exchange..:)

Single loops have one flow rate, that would have three individual flow rates, they just share the heat exchanger function is all.

Probably not a mesurable temp difference vs. one single loop with pumps in series, but it would minimize the maximum pressure in the system which could be helpful for really extreme pumping needs and thin walled tubing.

[Waterlogged]

Just to go against the grain, it's not a single loop...it's separate single pumping loops with a shared reservoir/heat exchange..:)

Single loops have one flow rate, that would have three individual flow rates, they just share the heat exchanger function is all.

Probably not a mesurable temp difference vs. one single loop with pumps in series, but it would minimize the maximum pressure in the system which could be helpful for really extreme pumping needs and thin walled tubing.

If your responding to my posts Martin, I'm getting the distinct impression the your actually going with the flow on this and I'm actually the "loon". Certainly wouldn't be the first time for me. ;) I stick by all my "ramblings" though. :nuts: