Q. about loop configuration. single vs double

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply

Should I run a single loop with two rads or two separate loops?

Single loop - the eheim pump will handle it, no worries
3
75%
two separate loops with two pumps and two reservoirs
1
25%
two loops with an alternate configuration - please advice
0
No votes
stick to air-cooling and leave WC to the experts you noob!
0
No votes
 
Total votes: 4

netrunner
Posts: 2
Joined: Mon Feb 22, 2010 5:28 am
Location: Oslo, Norway

Q. about loop configuration. single vs double

Post by netrunner » Mon Feb 22, 2010 5:49 am

My main goal for water cooling is to get a silent pc, but also I would like to be able to over clock, but maybe not to the extreme.

I am installing water-cooling in a Obsedian 800D and I have got Ek Supreme water-cooling-blocks for cpu (i7-920), mosfets, nb/sb and gpu (radeon 5970). I have got a Xchanger 360 rad and a AC "aquastream XT", 12V, Ultra pump and were planning to run in one loop, but have received feedback it will not be efficient cooling/flow.

I have then decided to either install two separate loops or put another 240 rad in before the gpu so i have f.eks res > pump > 240 radiator > CPU > Mosfets > NB/SB > 360 radiator > GPU > res. If I install second loop perhaps have cpu on one loop and chipset + gpu on the other.

My questions are:

1.will I have enough flow/pressure to run this loop with one eheim xt ultra with a 360 and a 240 rad, or is it recommended to install 2 separate loops instead.
2.if i install an extra loop will i have to use 2 reservoirs and 2 pumps? i have no previous experience in watercooling, so it might be a stupid question.

eit412
Patron of SPCR
Posts: 46
Joined: Tue Jul 24, 2007 6:53 am
Location: Texas

Post by eit412 » Mon Feb 22, 2010 10:47 am

I do not know much about water cooling, but i know that you will not need 2 reservoirs even if you do 2 loops. The tube coming out of the reservoir could be split to the two pumps with a Y connection. if you do two loops, I would not put anything on the 5970's loop as it will most likely be your hottest component.

If you want to have two pumps in line with each other it would not give nearly as good of performance as two loops, but they would be redundant, so if one failed you might have time before everything fries.

fpsrandy
Posts: 37
Joined: Fri Nov 06, 2009 10:25 am
Location: Winnipeg, MB, Canada

Post by fpsrandy » Tue Feb 23, 2010 10:54 am

Im not sure what an "ultra pump" is, but if you go get a laing mcp655 pump it should powerful enough for the setup you seem to be planning.

http://ncix.com/products/index.php?sku= ... e=Swiftech

AKA Swiftech MCP655, DangerDen D5, Laing MCP655.


If you are still concerned of the flow rate and pressure, I would ditch the reservoir and put a t-line in. t-lines can take awhile to bleed out, but I have always seemed to have better pressure. You don't need a reservoir if you have a t-line

if you need more pressure, go with 3/8 ID tubing, if you want better flow rate get 1/2 ID tubing.

spookmineer
Patron of SPCR
Posts: 749
Joined: Sat Nov 11, 2006 6:02 pm

Post by spookmineer » Tue Feb 23, 2010 3:13 pm

fpsrandy wrote:if you need more pressure, go with 3/8 ID tubing, if you want better flow rate get 1/2 ID tubing.
The pressure is best kept intact with 1/2 ID tubing (the resistance is less).

Assuming the difference in pressure loss between the two tubings is negligable, the flow rate will also be nearly the same for both tubings (although obviously the smaller diameter tubing will have faster liquid speed - the flow rate will be almost similar, because it's determined by pump specs).

1398342003
Posts: 324
Joined: Wed May 07, 2003 10:35 pm
Location: Surrey, B,C

Post by 1398342003 » Fri Feb 26, 2010 7:35 pm

If you want to maximize pressure use two pumps in parallel. A Y splitter, pumps A and B, and another splitter as a joiner.

1/2" pipe is better for high pressure and high flow, but with modern blocks the difference is small. Try to avoid tight bends and 90 degree elbow joints as they impede flow, but again the effect is negligible.

If you do get a Liang D5 pump/s make sure they are the kind that supports speed control. Pumps can be very noisy at full power, mine is at about half power to keep noise in check. Despite that, it's still one of the two loudest parts of my system.

I second the T line suggestion. When fluid enters a reservoir it slows down, losing it's energy, so the pump must re-accelerate the fluid. With a T line the fluid never slows down, so there is less work for the pump to do.

As far as I've seen watercooling the mosfets and sb/nb provides no real benefit, and will increase resistance in the system. My i5 750 runs at 4.2GHz @ 1.416v, along with a 9800GTX+, a Liang D5 at half power, and a Thermochill 120.3 radiator with three 120mm S-Flex fans at low speed. It's perfectly stable, and probably could go further.

PhilW
Posts: 27
Joined: Sat Mar 01, 2008 10:06 pm
Location: Singapore

Post by PhilW » Sat Feb 27, 2010 3:25 am

I've a MCP655 driving a CPU, NB, SB, GPU, 1 x 240mm rad and 2 x 120mm rads.

Phil

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Sat Feb 27, 2010 4:02 pm

1398342003 wrote:If you want to maximize pressure use two pumps in parallel.
If you want more pressure, putting the pumps in series would be better.

1398342003
Posts: 324
Joined: Wed May 07, 2003 10:35 pm
Location: Surrey, B,C

Post by 1398342003 » Sat Feb 27, 2010 6:17 pm

HammerSandwich wrote:
1398342003 wrote:If you want to maximize pressure use two pumps in parallel.
If you want more pressure, putting the pumps in series would be better.
Both parallel and series increase effective pressure, but you are correct, series potentially produces higher pressures. It really depends on the system's restrictiveness and overall pump power.

And system resistance isn't easy to measure overall.

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Mon Mar 01, 2010 5:57 am

1398342003 wrote:And system resistance isn't easy to measure overall.
That's certainly true. Very few sites produce good WC reviews.

Back to the OP...

Because you're looking for a silent PC, you should focus on the pump(s) & fans. If you have not bought the extra rad & pump yet, WAIT! Check your temperatures in a normal loop, testing with varying airflow & pump speeds. This is the information you'll need to make a good decision down the road.

netrunner
Posts: 2
Joined: Mon Feb 22, 2010 5:28 am
Location: Oslo, Norway

still waiting 4 parts

Post by netrunner » Tue Mar 02, 2010 4:02 pm

thanks for good advice

I have come to the conclusion that I will setup my loop as it is now with pump->3x120rad->cpu>mosfets->ek-fb->gpu(5970)->res and just get some experience on how my system works and how to measure what need be measured.

I am just waiting for two essential parts, fittings for my pump that should be sendt in an envelope from germany today, so by the weekend I hope to have everything i need to add water to my loop. at least it is looking good so far, and I got my powercolor 5970 the other day, damn it looks good and it is soo heavy!!

jonj678
Posts: 20
Joined: Mon Jun 22, 2009 12:28 am
Location: England

Post by jonj678 » Mon Mar 08, 2010 10:12 pm

There are some terrible lies posted in this thread. If you don't know what you're talking about, remain silent. Watercooling is not often intuitive and spreading misinformation helps noone.

You have a very restrictive loop, by any standard. Particularly the ek supreme, this performs brilliantly when flow rate through it is high but not very well when it isn't. It's also very restrictive. Open it up and remove the steel impingement plate, this will significantly improve flow rate and decrease temperatures in your case. A restrictive loop means flow rate will be low. This does not mean temperatures will be terrible, just that they won't be as good as they could be. You're aiming for quiet, not performance, so this is fine.

Two loops means two reservoirs, two pumps. One reservoir with two pumps drawing water from it behaves much the same, with some heat transfer between the two streams. Mixing in the reservoir can be quite limited so it is worth having a radiator as well as a pump in each line leaving the one reservoir.

An 18W laing ddc will pump water through this much resistance, but not at any great rate. I'm now using two 18W pumps on a similarly restrictive loop and seriously considering a third, adding the second took ten centigrade off cpu load temperatures so I was definitely flow rate limited, a rare thing for computer water cooling.

Changing tubing diameter to 1/2 will not help, neither will using a T line instead of a reservoir. If 1/2 doesn't lead to much longer lengths of tube (given he's watercooling his motherboard it surely will do) then it can offer marginal improvements, likewise T line vs reservoir. The decrease in resistance will be so negligible compared to the rest of the loop that it's ridiculous to suggest this as a solution.

Do not put pumps in parallel. The optimistic result of this would be doubling flow rate at the same pressure. Since the back pressure the pump is currently working against is so high all you would achieve is two pumps running hotter and pretty much the same flow rate. Pumps in series roughly doubles the pressure and works well with centrifugal pumps. If you try this with mismatched pumps you'll kill one of them. This would be a worthwhile consideration if the OP was more interested in performance than in noise.

Watercooling the motherboard provides benefit mainly in very low airflow conditions when overclocking, if at stock or with much air passing over the board it's not required. In the vast majority of cases it's not required, much like watercooling ram.

If you know the head pressure of the pump and the flowrate you can take a fair stab at the resistance of the system. In this case the majority of the resistance is coming from the ek supreme, with the gpu coming second if it's a full cover block.

I'm strongly in favour of series single loops over parallel. Parallel always has the potential to outperform series, but is also always more difficult to optimise. The standard of putting the graphics card in its own loop and letting it run hot because you don't care that it's at 70 degrees is quite sensible, but beyond thermal isolation of a particularly hot thing it becomes quite tricky. In this case however, where you want quiet, there's no point. Keep it simple.

Post Reply