Cooling Series or parallel water cooling?

I've got a question about how to route my tubing in my small bore (6mm ID/8mm OD) water cooling system.

I've seen a few other small bore systems that have everything ran in series, but they are all using something like the Eheim 1046 or 1048 pumps which are

1046 = 79GPH w/ 3'11" delivery head (1/2" input - 3/8" output)
1048 = 158GPH w/ 4'11" delivery head (1/2" input - 3/8" output)

I'll be using the Eheim 1260 which is 602GPH w/ 10'2" delivery head (1" input - 3/4" output)

I'll be pumping through the radiator and 4 water blocks (CPU, GPU, Chipset, and HDD). Because of the output of the pump I'm thinking it may be better to run all 4 blocks in parallel. I would run 1/2" ID tubing and split off four 1/4" ID tubes from that to go to each block. My only fear is that if the resistance to flow is not equal in all the blocks then some of them may be neglected adequate flow.

What do you guys think? Your opinions are very valuable to me.

Thanks!

 

Buy four pumps/rads..... Just kidding, I would check the flow rates after each of the four water blocks. Then limit the water flow to the higher flow waterblocks until they are all equal. Yes that will limit overall flow but not by much because it will just go one of the other four ways (take the easiest path).

-Fry

 

Okay first off need to know more about your setup. Are you going for low flow blocks to go with the small bore tubing.

In either case a 1260 is overkill. A 1046 would be fine for a small bore system and a 1250 would be fine for a high flow system. There are other alternatives like a Laing DDC etc.

Generally I say running things in series just makes life much easier.

 

For now I will be using all Koolance blocks, but I may convert to Aqua Computer blocks soon.

Another reason for the 1260 is because I'm setting up the pump, reservoir, and radiator in the basement below my office so I'll need to pump straight up 6 feet. The 1046 and 1048 don't have the delivery head to meet this requirement... the 1250 does, but I figured it's better to be safe than sorry and go overkill.

So... series or parallel?

 

As FRY said if you go parallel then unless the flow through all of the blocks is balanced you'll get good flow through one and poor through the other three as the water will try to take the pathe of least resistance. You ought to go in series that way al the water goes through all the blocks, no messin'

 

Don't you think that the CPU and 6800 Ultra and eventually a 2nd 6800 Ultra in SLI will add too much heat to the water if run in series?

I'm thinking that each time the water passes through a block in series it's going to get hotter... therefore reducing it's potential to cool the next block.

Is it not enough to worry about?

 

I just got a Sli set-up running in series which you can see by clicking the picture below to the prod log.

The flow goes into the bottom card first and then into the top one and the temp is around 36 on the top card and 40 on the bottom card which doesn't make sense but i'm guessing its because there is no air flow between the two cards and the chipset gets pretty hot which must effect the bottom card somehow!

Don't think a few degrees either way is going to make any real difference to the overall performance though.

I'd try and run in series if you can as running in parallel can sometimes lead to water not circulating in one (or more) of the hose and blocks..........usually the blocks with longer hose or greater resistance.

 

Coolmiester - Real nice setup.

If I were running 1/2"ID tubing like you I may be more apt to accept running in series because you're probably flowing a lot more water. Although personally I believe the pump, rad, and blocks are all more important that the tubing size.

Here's my new proposed idea...

I'm scrapping the HDD cooler.

Run 1/2"ID tubing from the pump upto a manifold that splits off into three individual 1/4"ID tubes. One for each waterblock in parallel (CPU, GPU, and Chipset). Then exit the waterblocks and run each individual 1/4"ID tube back to the reservoir each with it's own flow meter and adjustable valve; 3 flow meters and 3 valves. This way I could equalize the flow between each of the 3 circuits.

Here's a pic of the John Guest valve.


What do you think?

 

Here's my effort

As you can see, it's a bit heath robinson, but it does keep everything pretty cool. I have made some changes since I took the photographs and I will make some more soon. I have removed the elbows from the radiator, shortened the 25mm ID tubing and I currently have a Mobile Athlon XP2600 running at 12.5 x 206Mhz (2575Mhz) which remains around 13 celcius above ambient regardless of load the same with the GPU on the (also overclocked) graphics card; I've never seen either go over 37 celcius.

I have some Danger Den blocks and barbs with 1/2" ID tubing to replace the Asetek blocks with push fit connectors, which cause me problems, and the 8mm ID tubing. I also want to tidy the whole thing up a bit, I have the parts, I just need the time.

There is no way of balancing flow between the blocks on my set-up. My thinking was to use large bore tubing, try and keep flow restriction to a minimum throughout the system and use a big pump - it seems to work.

 

I can tell you right now you're gonna be twiddling with those valves a heck of a lot before you even get close to what you want, and even then you won't get exactly even flow through all three circuits. BUT, it might work to your advantage, since you want more water to flow to your hot components (CPU, GPU) and less to your not so hot/not so heat critical components (hard disks). But again, it's gonna require a lot of twiddling before you get things the way you want them. Also, valves restrict flow (that's what they do). You also run into a ton of mucky stuff like what happens to water viscosity as temperature changes (i.e. load vs. idle). In other words, even though you might have the flow rates right for one instance (i.e. idle), they'll change over the course of the day due to load, etc.

If you still wanna go the manifold/valve setup McMaster sells a sexy looking 3-way manifold that you might find useful
but alas it's alu and it's for air. But I bet someone has something similar made for water (just anything but those butt-ugly block manifolds).

Also , I THINK I remember seeing manifolds with built in restriction valves, might look around for those and it'll be a lot cleaner looking than having three 2-way valves floating aorund the case.

Anyways, have fun with whatever you choose to do, and if it doesn't work then it's all good cuz you know you can always fall back on a series loop.

-L6

 

That's what I was thinking... It'd be a easy no brainer to fall back to a series setup.

I know what you mean about playing w/ the valves to get the flow perfect, but like you said, as long as I have good flow to the CPU and GPU I should be fine. The valves will just be a way to kind of keep things in check. My luck I'd get the system up and running and 90% of my flow would be going to the chipset cooler and the other 10% shared between the CPU and GPU.

My main objective here is a little overclock with SILENCE!!! I'll have my reservoir, pump, and radiator in the basement below my office so I won't even hear the noise of the radiator fans... even if I crank them up. I have a finished basement, but the mechanical room where my hot water tank and furnace are just happened to be directly below my office. This room is insulated itself (walls and ceiling) so you can't hear the furnace run anywhere in the house. Maybe I'll put three of those 200+ CFM Delta 120mm fans on my radiator.

I can't wait for silence. The chipset fan on my A8N was loud enough before it started to take a crap. It's been grinding and whining for about a week now. It's going to die any moment.

 

If you're after complete silence I would maybe suggest a low flow loop to go with the other one. You will then be able to watercool your PSU and HDD. The HDD is going to be the loudest thing in your case but in a sound proof HDD cooler it should be silent.

Some sound proof matting and rubber grommets will also help reduce the noise.

 

All your worries about flow seem a bit misguided to me. You are concerned enough to think about using a parrallel setup to keep the flow as high as possible yet your using koolance blocks in a 1/2" set up. Soon as the water hits the first block its restriced, regardless of parrallel or series the restriction is still there and is the same. As coolermeister has pointed out you also run the risk of having dead spots where no water will flow at all. Series works out better in nearly all situations and you would probable be best to rethink your choice of blocks or pump as they dont seem suited to each other.

Measuring pump performance by head pressure can be a bit misleading too. Pug, the owner of www.wizd.co.uk, set up a 1048 or 1046 in his garden and had it pumping water out of his first floor window (with out restrictions) which is a lot higher than 5 foot. if we are lucky he might post the pics!

 

I don't necessarily agree with that. I'm using 1/2"ID from the pump and then splitting off into 4 individual 1/4"ID circuits. The area of a 1/2" circle is almost exactly the same as the area of four 1/4"ID circles.

Area of 1/2" circle = 0.1963 sq.in.
Area of 1/4" circle = 0.0491 sq.in.
0.0492 x 4 = 0.1964

Don't think of 1/2" as being twice the size of 1/4". If you do the math it's actually 4 times bigger. That's why i'm going from one 1/2" to four 1/4". Makes sense to me.

In regaurd to the 1048. Eheim specs. say the delivery head is 4'11" and according to a rep. their head specs. are conservative by approximately 10%. This would put the actual delivery head of the 1048 right about 5'5" which is right on par w/ Pug's test. Now if he was accomplishing this with the 1046 then I have no explanation.

The reason I'm using the 1260 is because I'm pumping straight up 6 feet and then have to overcome the resistance of 4 waterblocks and a radiator. The pump seems properly suited IMO w/ a delivery head rating of 10'+.

Am I figuring this wrong?

 

With regards to the loop, if you were going paralell then to get the right kind of performance in every block your probably going to need a small pump in each paralell part. If you think about what your doing with the valves your not increasing flow to other tubes, juts resrticting flow of the less restrictive parts of the loop. Not worth the hassle of setting up IMO. Again i say low flow, lots of blocks in series. This is really where low flow shines.

With regards to pump head in a closed loop it doesnt matter much. It can be thought of as the ability to push past restrictions. Pug did a small selection of tests on a 1048 and an L20. Here is a link to his thread on wizd. Pug says "Pump!", Eheim says "How high?"

[edit] should you try the paralell route though Good luck. Would be a fun project.[/edit]

 

I agree that you are restricting the circuit that your closing the valve on, but as you do that your are increasing the flow to the other circuits. It doesn't happen in linear proportion, but it does happen

Turn on your bathroom sink to full flow. Have someone else turn on the kitchen sink. When the kitchen sink gets turned on the bathroom flow clearly is reduced and then increased when the kitchen goes off.

Why wouldn't it be the same w/ a computer coolin system with 4 circuits. Close one valve in a 4 circuit system and flow will slightly increase to the other 3. And we're talking about barely closing the valve here... I think some people are getting the idea the valve is going to be closed half way or more. It's probably only going to have to be closed less than 10%.

Another thing. I really think that all the flow rates will be nearly equal anyways. I'm using all Koolance blocks and they're all VERY similar in design and size and they'll all have equal length tubing ran to them.

After all this discussion I'm going to have to run the system in series and parallel and compare temps on all the blocks just to satisfy myself. I'll post pics. and results. Does anyone know of a good site to host pictures for free?

 

I completely understand what your saying about flow changing in different parts of the loop, i guess i should have put that in my post, Im just not sure its worth the effort to try and set up effectively. Do you have any way of measuring the flow in each part of the loop?

Excellent

If you dont find anyone reliable i can host the pics for you. PM me if you need to

 

WOW!!!

I got my pump today. I really don't think I even have to worry about any of the 4 circuits being deprived of water flow. I hooked that sucker up and it really pumps out some water!!! Four 1/4"ID circuits will definately be restricting this pump.

What are the PSI ratings on the Koolance blocks?

 

I think he said he was going to use a flowmeter for each loop.

To Bajastx: Isn't a mech/furnace room kinda hot?

-Lemur 6

 

I thought about that. I turned the furnace to high and let it run for 10 minutes, which it never normally does and the ambient temp in the room only went up less than 1.5°F.