Overclock.net banner

1 - 20 of 23 Posts

·
Registered
Joined
·
2 Posts
Discussion Starter #1
Hey guys, im putting together a mini-itx loop and running into size constraints, (surprise surprise)
I havent actually seen or heard it done before but im assuming splitting the loop flow to two different blocks and radiators would work just fine with the use of t-splitters on the pump out and res in?

Feed back would be appreciated thanks!
 

·
Newsfiend
Joined
·
2,421 Posts
Between the bigger rad and CPU blocks being more restrictive in general not to mention the T line right in front of the pump I have a feeling the CPU won't be getting much in the way of flow. If this is just for aesthetics then I wouldn't even try it.
 

·
Overclocker in training
Joined
·
13,920 Posts
Hi,
Depends on which cpu platform really
Single pump on a cpu and gpu block is hit and miss pressure wise if it works out well or not that I've found out
Many will separate the two in different loops and 2 pumps for best cooling because pumps bite for both in a single loop.

Single reservoir and 2 pumps might be workout better :/
 

·
Totally Tubular
Joined
·
1,508 Posts
There would be no benefit to running it that way. The coolant will flow where there is the least resistance, so one half of your loop would always get less flow.

Better to make it all one loop in series. Then the loop will equalize in temperature and loop order doesn't matter.

So:

Reservoir > Pump > Big Rad > CPU > GPU > Small Rad > Reservoir
OR
Reservoir > Pump > Big Rad > Small Rad > CPU > GPU > Reservoir

Could also run CPU/GPU before the rads. Whichever is better/easier for tubing runs.
 

·
Registered
Joined
·
184 Posts
Like most said, it be a bad idea as flow rate will be divided by half, the least resistance pathway will get the most flow. If you really want to go in that direction, like been suggested 1 reservoir & 2 pumps. If space a dual loop would be much better looking.
 

·
Banned
Joined
·
3,726 Posts
Hey guys, im putting together a mini-itx loop and running into size constraints, (surprise surprise)
I havent actually seen or heard it done before but im assuming splitting the loop flow to two different blocks and radiators would work just fine with the use of t-splitters on the pump out and res in?

Feed back would be appreciated thanks!

Visual Reference courtesy of my degree in graphic design.
well that is how I build all of my water loops.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html
http://www.overclock.net/forum/61-water-cooling/1615072-cpu-radiator-upgrade-water-cooled-rig.html
http://www.overclock.net/forum/62-peltiers-tec/1651377-full-system-tec-cooling.html

so yes it will function fine in that configuration. I would suggest two pumps, maybe a duel pump top, but even with one pump that configuration will function well.
 

·
Registered
Joined
·
109 Posts
My opinion, Res>pump>rad> GPU>rad>cpu>res. That way they both get fresh cool water from a rad. But I don't believe your way is that bad. Flow loss wont be anywhere near as bad as some would think. I've used Ts without any issues.
 

·
New001
Joined
·
2,668 Posts
Visual Reference courtesy of my degree in graphic design.
Haha love it

My opinion, Res>pump>rad> GPU>rad>cpu>res. That way they both get fresh cool water from a rad. But I don't believe your way is that bad. Flow loss wont be anywhere near as bad as some would think. I've used Ts without any issues.
By increasing the surface are of tubing (almost doubling) with two fully parallel loops, wouldn't you INCREASE flow, rather than have loss? I'd say his loop should be running as much as 33% faster than a single/inline loop... will confirm with math later...
 

·
Banned
Joined
·
3,726 Posts
Haha love it



By increasing the surface are of tubing (almost doubling) with two fully parallel loops, wouldn't you INCREASE flow, rather than have loss? I'd say his loop should be running as much as 33% faster than a single/inline loop... will confirm with math later...
you are correct!

no math needed, here is a live example of a water loop that has both the components, and the radiators in parallel. a serial water loop of my hardware and one pump, runs at 2.4LPM. the parallel water loop, of the same hardware, and only one pump runs at 6.8LPM.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html
 

·
Registered
Joined
·
109 Posts
Haha love it



By increasing the surface are of tubing (almost doubling) with two fully parallel loops, wouldn't you INCREASE flow, rather than have loss? I'd say his loop should be running as much as 33% faster than a single/inline loop... will confirm with math later...
I have no idea what you are saying. It's just gobbledegook. I'm not sure if you are criticising him or me, you seem to be having an each way bet. :)
 

·
Registered
Joined
·
1,165 Posts
The plan you have is just a simple parallel loop and it will work fine with the exception that the pump has to work harder with the higher system flow rate.

Unless both branches of the loop are exactly equal in restriction then they won't get exactly equal flow but that is not the problem that some believe. For one branch to get 50% more flow eg; 4Lpm and 6Lpm, or 2Lpm and 3Lpm, one branch has to be twice the restriction of the other. Like say two blocks on one and a single on the other. For one branch to get only half the flow of the other requires 4 times the restriction. You can achieve those sort of differences by say having the rads on one path and the blocks on the other but with them roughly equally seperated with one block and rad each the difference in the paths you have drawn is far less than either of those examples unless your blocks are unusual or old, or one rad exceptionally high restriction, so the flow rates wont be very different at all most likely.

Parallel loops present much lower resistance to flow to the pump than series loops so the flow rate will be comparatively very high which makes the pump work harder, use more power and run hotter.
Although the system flow rate is higher it has to be split between two paths so its quite normal that the flow rate the blocks and rads receive is actually lower than in a straight series system. Thats no great issue though. If you want to try it out then go for it without any concern.
 

·
Newsfiend
Joined
·
2,421 Posts
Thanks for chiming in Ashcroft my original comment was too harsh. As with a lot of things in watercooling it all depends on your goals.
 

·
Registered
Joined
·
599 Posts
It will work as long as the flow in the respective loops is still good. Your cpu loop if using a block with a jetplate will care a lot more about this. As stated already though there is no reason to do this other than you want it to look a particular way. Cooling will be worse than if you had everything in serial.

My opinion, Res>pump>rad> GPU>rad>cpu>res. That way they both get fresh cool water from a rad. But I don't believe your way is that bad. Flow loss wont be anywhere near as bad as some would think. I've used Ts without any issues.
The loop order does not matter. You just want the res before the pump so it does not run dry. Other than res before pump you can place everything else in any order and it will not matter. The coolant in the lines will have the same temperature thought the loop. Now it is true that it can take some time for all the coolant to equalize in temperature in massive loops but the results are the same in that in time all the coolant will be the same temperature.

Coolant is in a closed system, it is not like air cooling where blowing hot air from your gpu on your cpu would matter.
 

·
Banned
Joined
·
3,726 Posts
It will work as long as the flow in the respective loops is still good. Your cpu loop if using a block with a jetplate will care a lot more about this. As stated already though there is no reason to do this other than you want it to look a particular way. Cooling will be worse than if you had everything in serial.
this is a common misconception, the main reason anyone would have for going from a simple serial water loop, to a complicated parallel water loop is to lower temperatures on the system. when done correctly, a parallel water loop, will always be cooler than a serial water loop, of the same hardware.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html

my component temperatures went down, after paralleling the water loop.
 

·
Unwilling Beta tester
Joined
·
7,188 Posts
@jtblache - that will work, although how well it works depends on your particular setup. Be aware that although your overall flow rate will go up, the higher restriction side of your loop (typically the CPU side) will get less flow than the other side - you will need to monitor things carefully on your first run. For splitting loops a Y splitter is better than a T-piece if you have the space.

Can I ask why you are planning on doing it this way?



@toolmaker03 - your proof for parallel loops always being better than serial seems to be based on the fact that you made rearranged your existing loop and got better temperatures. However you also added a second pump to your parallel loop whereas the serial loop ran on one pump. I would argue that this is not a true test of the two setups.





There is also a lot of misconception going on in this thread with regards to water temperatures around your loop.

Misconception: Water temperature will be the same around your loop

No, it won't. It will obviously be warmer after absorbing heat from a hot component than it will be after dumping heat in a radiator. If the water did not change temperature it would not be able to transport heat away from your components. The reason why loop order often doesn't matter is because the differences in temperature are fairly small (more on this later).


Misconception: 'Equilibrium' means your loop is the same everywhere.

When your loop is in equilibrium it simply means that the same amount of energy is going in as is coming out. It does not necessarily mean it is the same temperature everywhere. A better description might be 'stable' or 'steady state'.


Misconception: Other than res->pump, loop order never matters

To be fair, this is true the vast majority of the time. Because water can absorb a lot of heat with very little temperature change (4.18 kJ/(kg.°C)), so as long as your flow rate is reasonable the temperature will not vary much around the loop. For reference, at 1GPM (4.55LPM) it takes 316W of heat to raise the water temperature by 1°C.

However in some cases it can matter. For example if you have multiple power hungry GPUs (think quad SLI or a mining rig), or a particularly low flow rate, or you are chasing every degree on your CPU, you would want to put your CPU before your GPU.
 

·
Banned
Joined
·
3,726 Posts
@jtblache

@toolmaker03 - your proof for parallel loops always being better than serial seems to be based on the fact that you made rearranged your existing loop and got better temperatures. However you also added a second pump to your parallel loop whereas the serial loop ran on one pump. I would argue that this is not a true test of the two setups.
well I would request that you reread that thread, serial vs parallel, as I did do a setup of a parallel water loop with only one pump, and it is part of the first post of the thread.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html

also I am not basing my opinion on just one build, I have three.
http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html
http://www.overclock.net/forum/61-water-cooling/1615072-cpu-radiator-upgrade-water-cooled-rig.html
http://www.overclock.net/forum/62-peltiers-tec/1651377-full-system-tec-cooling.html
 

·
Unwilling Beta tester
Joined
·
7,188 Posts
well I would request that you reread that thread, serial vs parallel, as I did do a setup of a parallel water loop with only one pump, and it is part of the first post of the thread.

http://www.overclock.net/forum/61-water-cooling/1573189-serial-vs-parallel-9-6lpm.html
Right - now I see it.

You had a flow rate of only 2.4LPM with the serial loop and one pump? Something is very wrong about that. No wonder you had relatively poor temperatures.


And did you build a series version of each one of them to compare it? And do any of them remotely resemble the OP's loop?
 

·
Banned
Joined
·
3,726 Posts
Right - now I see it.

You had a flow rate of only 2.4LPM with the serial loop and one pump? Something is very wrong about that. No wonder you had relatively poor temperatures.

And did you build a series version of each one of them to compare it? And do any of them remotely resemble the OP's loop?
well in my opinion, all of my parallel water loops are similar to the parallel water loop that OP plans to build. the first and second builds started out as serial water loops. I upgraded the serial water loops to parallel water loops to improve system temperatures. my TEC build has been parallel from the start.

now if you will go further down in the serial vs parallel thread you will see that the most restrictive part of my system is the GPU water blocks with a flow rate through them of 3.4LPM. now regardless of what chart you may be looking at, the way I find out how restrictive a component on my water loop may be, is by actually testing it myself. so while in your mind you think there is something wrong with the water loop. I know that it is functioning as best it can, because I have tested all my components, and have connected them to a serial water loop one by one. that is how I know that 2.4LPM is exactly what the final outcome of those components, in a serial connection, and one pump will equate.
 

·
Overclocker in training
Joined
·
13,920 Posts
Hi,
lol I believe the op left the building :D
 
1 - 20 of 23 Posts
Top