PDA

View Full Version : 2 loops or pumps in series?



scottc19
09-10-2008, 08:17 PM
I am deciding between 2 loops or 2 pumps in series, originally I thought 2 loops would be better, but since I have ample cooling 5 radiators it seems that twice the flow will ultimately give lower temps. cpu, gpu, nb in loop. e8600 is cpu. 5 rads = 2 double and 3 single. Pumps are mcp355 w/alphacool top and mcp350 w/oclabs top. Can anyone confirm that series should be better than 2 loops?

scottc19
09-10-2008, 08:32 PM
mcp120, BIXI, BIXII, pa120.2, BIP, dtek fuzion v1, mcw60, mcw30, all tubing is 7/16, so basically 1/2. Pumps are mcp 355 w/alpha cool top, and mcp350 w/oclabs top

SNiiPE_DoGG
09-10-2008, 08:36 PM
you can not use 2 pumps of varying speeds in one loop, they must be the EXACT same pump.

scottc19
09-10-2008, 08:39 PM
Ok so then I could mod my 355 to a 350. So how about two 350's in series or a 355 and a350 in 2 loops, rads could be in any combination. 1 loop would be easier for me as I would only have to buy 1 reservoir.... Heres my worklog so you guys can get an idea of what I'm dealing with http://www.xtremesystems.org/forums/showthread.php?t=200924

SNiiPE_DoGG
09-10-2008, 08:54 PM
I do believe one can easily mod a MCP350, into a MCP355 ....... :up:

true, but I was only speaking in the context of 355 matched with 350.

essentially the pumps must be of the same model series and run at the same RPMs

D5v@4 = D5fixed

mcp355 = modded mcp 350 :up: :up:

pika198
09-10-2008, 09:09 PM
the mod to make mcp 350 to 355 is just a simple solder connection :D

scottc19
09-10-2008, 09:44 PM
My 355 can be modded back in forth, but my 350 is fixed, its the older version...

shabranigdo
09-10-2008, 11:45 PM
I would suggest 2 loops. The processor is the most temperature critical component so you don't want to add any heat from other sources if you can. Things like mosfet NB/SB and GPUs have alot more tolerance to heat. I would put the CPU alone with enough radiator for its heat output and put the 30 and 60 in a second loop. That said if both pumps are the same one loop will give you protection if one pump fails, though other hardware and software solutions can shutdown your system on pump failure. There are some stickyed threads that will help you calculate your heat output of components and heat dissapation for given radiator and fan combos. That will let you calculate how much radiator you need for your specific setup.

twwen2
09-11-2008, 12:16 AM
Me too. Pumps in series offer diminishing returns (from the testing we've seen here recently), whereas splitting loops enables you to isolate significant heat sources (e.g. CPU from GPU etc).

Dual-loop is the way to go i reckon.

Snyxxx
09-11-2008, 09:29 AM
Agreed.

I just performed a test with one pump (DDC-2 With Orange Impeller) and two radiators with a bunch of blocks (CPU, NB, SB, 2 Mosfets and 2 X FC 8800 FC SLI) .

I measured all my temperatures and then simply added a second identical pump in series.

I was a little shocked that my temperature only dropped by ~1°C on everything.

For me, two loops would offer better performance just as others above have stated.

But you should experiment yourself and see what is best for you. It is fun as well.

shabranigdo
09-11-2008, 09:42 AM
Another thing I forgot to mention, recently some of our testing community have found that by having 2 pumps in your loop the heat dump from the pumps is more noticable and can counteract the benefits of extra flow. I think it was a supprise to alot of people how much that actually effected your final temperatures.

F1ZZY
09-11-2008, 06:20 PM
you can not use 2 pumps of varying speeds in one loop, they must be the EXACT same pump.

Moot point now this late in the thread, but I believe the above not to be true.

As long as neither of the pumps max flow rate will be exceeded then mismatched is probably ok. (It will be almost impossible to exceed max flow rate in a loop with so much restriction too)

MoreGooder
09-12-2008, 03:16 AM
I've browsed thru this thread. Let me share a bit about my loop:

I use two pumps in my loop, but the loop has everything in series.

Res - pump 1 - 120.3 - cpu - nb - pump 2 - 120.2 - gfx - res.


The pressure in the loop is enormous. I don't have a way to measure the flow rate, but suffice it to say that I wish the res was about twice as large because the torrential flow makes it difficult to bleed out the air bubbles when I replace the coolant.

I do understand the value of two loops. However, in my opinion, the simplicity of a single loop and the decreased amount of tubing is a strong selling point to this method. More importantly, if one of my pumps fails, the other pump will keep the water flowing albeit with less flow rate. As it is now, when I blow the dust bunnies out of my case I make a special effort to put my fingers on the pumps to see that they are both still vibrating as an indication that they are still pumping.

With regards to the extra heat dumped by having two pumps: The way I see it I would get roughly the same amount of heat dumped with a higher capacity single pump compared to two pumps, but without the added security that the two pump approach provides. Plus, I have plenty of excess cooling capacity with the two radiators anyway.

Hope this helps your decision. Granted, I don't consider myself an expert, but I've been water cooling and reading these forums for many years (but didnt register until recently). Keep in mind that I consider the piece of mind factor of having two pumps in series an overwhelming reason for this approach. See my sig for more component details.

MG

Bobly
09-12-2008, 03:41 AM
Well there's a + and - for either option:

With 2 loops, you can separate the 50-80C components who require cooling but don't mind running warm (GPU) from the 30-50 components who really like the cold (CPU).

With 1 loop and 2 pumps you have redundancy, if one pump fails the other will still be around to pump the coolant through the system.

hotdun
09-12-2008, 08:04 AM
you can not use 2 pumps of varying speeds in one loop, they must be the EXACT same pump.

Is this necessarily true? What if the loop looked like this:

RES - PUMP1 - RAD1 - CPUBlock - NB Block - PUMP2 - GPU Block - RAMBlock - RAD2

Wouldn't the restriction of the blocks slow the flow enough to where different pumps with different flow rates/head would be beneficial and not cause any issues?


I'll be running Dual-MCP355's, one with XSPC Reservoir top and one with XSPC top, in CPU only EK Supreme loop. Will also be running dual PA120.3's in said loop...:up:

You got an ETA on this?

IanY
09-12-2008, 08:12 AM
I think the pump heat dump concept is sometimes misunderstood.

Ceteris Paribus, with the same blocks and one radiator, adding a second pump introduces more heat to a loop and the water temps go up. How much it heats up is dependent on the pump's heat dump. Naturally, an Iwaki RD30 will introduce more heat to the loop than a 10watt DDC.

If you add a second pump and a second radiator to the loop, then I would be willing to bet the house that the added cooling capacity from the second radiator will far surpass whatever pump heat dump there is from a second pump.

Look at RRR. he's added a second pump, and a second PA120.3. I would bet big time that his temps from the added flow and added radiator capacity will improve from using one pump and just one radiator.

IanY
09-12-2008, 08:24 AM
Andy is wise.

Seperate loops are easier to maintain.

hotdun
09-12-2008, 08:44 AM
you can not use 2 pumps of varying speeds in one loop, they must be the EXACT same pump.

Is this necessarily true? What if the loop looked like this:

RES - PUMP1 - RAD1 - CPUBlock - NB Block - PUMP2 - GPU Block - RAMBlock - RAD2

Wouldn't the restriction of the blocks slow the flow enough to where different pumps with different flow rates/head would be beneficial and not cause any issues?

What about the question on using 2 different pumps in the same loop.....say 1 D5 and 1 DDC?

hotdun
09-12-2008, 10:54 AM
...I wouldn't push it with 2 dissimilar pumps. Just seems like the unequal head pressure from one pump would cause to much stress on the other one.....


I agree...but only if the pumps were directly connected would this be a concern. Pressure and flow varies throughout the loop, right? So what if the different pumps were spaced out throughout the loop?

hotdun
09-12-2008, 11:56 AM
I see what your saying, but like I said. I Can't/won't prove it either way. Be worth a try though, let me know how it works out for you.

andyc

I'm gonna try the following and I'll let you know if anything happens:

RES - D5 - RAD - CPU - NB - DDC