2 pumps in one loop?

F

FatboyHK

Limp Gawd
Joined
Nov 14, 2006
Messages
268
Is it possible? I think yes, but does it make sense? I don't know.... Is it a sensible way to add reliability to my WC?
 
will increase head pressure for powering through multiple blocks, allows for dual loops later, I don't think the two pumps will make the system more reliable. But with two pumps if one fails the second could maintain the loop for at least long enough for you to shut down the system.
 
yea when one fails you have a few minuits to a few hours to shut down before the huge increase in restriction the dead pump adds burns out the other pump as well. but atleast the water will still be flowing for a little while. if your using two pumps, make sure they are the same kind and flow rate.
 
What if you ran both pumps in parallel straight from the reservoir? Have two lines from the reservoir go to the two pumps, then connect the pump outputs with a Y adapter. If one pump dies, would this not prevent the problem of flow restriction due to the dead pump?
 
Two pumps will increase max head and give you redundancy. The loss of a pump won't restrict flow too much. I've heard of people losing one DDC-2 and still have great temps with only one pump. It's just considered as another block in the loop. Just make sure you use identical pumps.

Never ever, put pumps in parallel. The only thing that should ever be placed in parallel are identical rads.
 
oh, so if I really wanna do this I need to arrange them in serial instead of parallel? Oh I thought it is smart to make them in parallel, thx for setting me straight :)

but may I ask why?
 
There was a huge debate about this on the XS forums and the math/physics nerds set it all straight w/ their theory and half a dozen members proved it with 1st hand experiments. I'll see if I dig up the link and paste.
 
the problem with parallel is that you would have to increse the size of the tubing at the Y fitting to keep the flow at reasonable levels. Also you need check valves to keep the two pumps from pumping against one another (too much restriction). Fluid follows the path of least resistance just like everything else, if the pump died the other would pump around the Y and back into the reservoir through the dead pump. Like Ranker said, two identical Rads or maybe blocks is all you want to run in parallel and only if the tubing feeding each is the same after the split, else you will get uneven flow or even no flow on one side of the parallel system.
 
Arcygenical said:
It's pretty much like dividing by zero. The world ends.
Click to expand...

Oh, and here I thought it just equaled "E" -glares at calculator-

So really the only way two pumps could possibly work in parallel was if there was some sort of one-way valve system setup to prevent one from returning through the other should one die. I kinda assumed since they were both applying pressure in the same direction, and each would be powerful enough to push the entire loop, the path of least resistance would have to be the remainder of the WC loop.
 
MjrStryker said:
So really the only way two pumps could possibly work in parallel was if there was some sort of one-way valve system setup to prevent one from returning through the other should one die. I kinda assumed since they were both applying pressure in the same direction, and each would be powerful enough to push the entire loop, the path of least resistance would have to be the remainder of the WC loop.
Click to expand...

Not true:

parallelChart.jpg


source
 
MjrStryker said:
So really the only way two pumps could possibly work in parallel was if there was some sort of one-way valve system setup to prevent one from returning through the other should one die. I kinda assumed since they were both applying pressure in the same direction, and each would be powerful enough to push the entire loop, the path of least resistance would have to be the remainder of the WC loop.
Click to expand...

well, both pumps are putting out positive pressure, it's just that one pump will probably flow a bit more water than the other one.

here's how it breaks down:
two in series: higher pressure, less flow rate drop with more resistance
two in parallel: higher flow rate, but more flow rate drop with more resistance

but with two in parallel, if just one dies, the entire loop is basically down

however, on the other side of the coin, it's perfectly fine to idle in windows without a pump running, i've tried it myself, and with a tdx, things only get to around 50c before leveling off, and it takes quite a long time to get there too :p
 
(cf)Eclipse said:
it takes quite a long time to get there too :p
Click to expand...

Yay for convection...

I run dual pumps at work all the time in parallel sometimes as many as 50 in parallel but again each one has a one way valve on it and they are feeding into a larger line than what they are pumping on. Of course these are huge pumps that generate upwards of 1200HHP each :D

If there was a way to get two pumps 1/2" ID into a larger tube, say 3/4" ID with a very effective Y fitting and some low pressure drop one way valves I bet you could nearly double the flow rate and keep the pressure constant
 
the only way you will find out is by experimenting & taking measurements...
 
however, on the other side of the coin, it's perfectly fine to idle in windows without a pump running, i've tried it myself, and with a tdx, things only get to around 50c before leveling off, and it takes quite a long time to get there too
Click to expand...

Yeah, I frequently boot up without my water lines attached... just the blocks.

Reassures that I havn't killed the motherboard somehow :p
 
FatboyHK said:
Is it possible? I think yes, but does it make sense? I don't know.... Is it a sensible way to add reliability to my WC?
Click to expand...

What kind of loop are you running?
What two pumps?

I don't think you would see much improvement.
If anything, you would dual radiator with one pump.
 
I have 2 MCP350 with modified Alphacool tops in series.
Loop= 1st MCP350 pump with Alphacool top & reservoir ->1st MCR200 Rad -> Apogee GT block -> DD 680i chipset block -> 2nd MCP350 pump ->2nd MCR200 Rad -> DD 8800GTX block -> Alphacool NexXxoS NB-SLI 1block ->back to 1st MCP350

CPU = QX6700
GPU = XFX 8800 GTX
Motherboard = EVGA 680i

Temps with 1 MCP350 on:
CPU = 37C, GPU= 47C, NB= 31C

Temps with 2 MCP350 on:
CPU = 35C, GPU= 45C, NB= 29C

With 2 pumps running and more noise, there was only a 2C improvement, so I unplugged the 2nd pump, and currently only using 1 pump in the loop.
 
^^ exactly what I did.

I have two Petra's Top DDC pumps in series and the temperature difference is far to low to justify running dual pumps. (btw , my loop was similar to his as well, except with an additional GTX)
 
Dual pumps will only increase performance substantially if you have restrictive radiators, waterblocks or many waterblocks in one loop.

With unrestrictive radiators such as a Thermochill PA's or Swiftech MCR's, you won't get too drastic of an improvement. However, that's a good thing as it means your components are efficient and top notch.

Two pumps in series is done for restrictive loops (crappy waterblocks/radiators) or most often for redundancy.
 
ShiningHolden said:
What kind of loop are you running?
What two pumps?

I don't think you would see much improvement.
If anything, you would dual radiator with one pump.
Click to expand...

the loop will include an Apogee GTX, 2x EK 8800GTX blocks, and a BI GTX 480 rad.
 
I vote two in series for that, or better still two loops with the CPU on a Swiftec 120x2 and the GPUs on a Swiftec 120x3 or if you can swing the big money get some PAs
 
FatboyHK said:
the loop will include an Apogee GTX, 2x EK 8800GTX blocks, and a BI GTX 480 rad.
Click to expand...

If you want top notch performance, put the GTX in its own loop. FC blocks increase temps by a large margin on the CPU due to the heat dump from the ram chips.
 
Ranker said:
FC blocks increase temps by a large margin on the CPU due to the heat dump from the ram chips.
Click to expand...

due to the heat from the ram? just how large of a margin are you talking about here? even overconservative calculations put the temp increase at 2-3C by adding the heat from the ram... most likely closer to 1C though.
 
(cf)Eclipse said:
due to the heat from the ram? just how large of a margin are you talking about here? even overconservative calculations put the temp increase at 2-3C by adding the heat from the ram... most likely closer to 1C though.
Click to expand...

I think he was talking about restriction? :confused: I'm not sure how much heat those little memory chips can put out either.. probly not much.

Though adding two EK 8800GTX full cover blocks to my loop heated up my CPU by 6 degrees.
 
(cf)Eclipse said:
due to the heat from the ram? just how large of a margin are you talking about here? even overconservative calculations put the temp increase at 2-3C by adding the heat from the ram... most likely closer to 1C though.
Click to expand...

The EK FC blocks are restrictive and the RAM dumps a large amount of heat into the loop. His CPU temps will rise significantly due to the restriction and the extra heat dump.

Within a single loop, going from an EK FC block on my SLI 8800GTX's to a D-Tek GFX, my GPU temps dropped by 6C and my CPU temps dropped by 8C. I saw similar results with the MCW60's. Everything else within the system was kept constant.

I love the EK FC blocks. But if you use them, make sure you're using them in a separate loop. If not, then don't expect a large amount of headroom in OC'ing your CPU. But if quiet is what your after, then don't worry about it at all.
 
Ranker said:
I love the EK FC blocks. But if you use them, make sure you're using them in a separate loop. If not, then don't expect a large amount of headroom in OC'ing your CPU.
Click to expand...

Well, my setup handled up to about 1.6 vcore... I left it at 1.55v for 3.8 GHz. It takes a hell of a lot of heat to fill up a PA120.3
 
(cf)Eclipse said:
due to the heat from the ram? just how large of a margin are you talking about here? even overconservative calculations put the temp increase at 2-3C by adding the heat from the ram... most likely closer to 1C though.
Click to expand...

The EK block increased my GPU temps by about 5c (I think it's due to contact and block design, not RAM heat) and my CPU temps didn't change at all. This was compared to an MCW60.
 
My EK block added about 5c as well.
I am going to run two loops, maybe even 3, when I SLI 8800GTX's.
 
FatboyHK said:
then a 6 degree increase is very reasonable
Click to expand...

6C increase on the video cards is reasonable since one isn't really temp limited in overclocking 8800's. They're voltage limited and one needs to do a voltage mod to really do some overclocking. However, the 8C increase in CPU temps isn't reasonable imo. My temps being at 55C or 63C makes it clear what's more favorable in terms of overclocking.
 
oh, I said it is reasonable because he went from a CPU only loop to a CPU + GFX + GFX loop without adding any pump, rads, or fans, and with that a temperature increase is very well an expected consequence.
 
Ranker said:
The EK FC blocks are restrictive and the RAM dumps a large amount of heat into the loop. His CPU temps will rise significantly due to the restriction and the extra heat dump.
Click to expand...

the restriction will definitly change things a lot
the heat from the ram, not so much. :p
 
You must log in or register to reply here.
Back
Top