Whats the best tube diameter?

[Crosstrack16]

Some say 3/8, some say 1/2?

[Cathar]

It depends. For blocks that have an inlet above the core and don't artificially restrict the inlet further (for example - the White Water block artificially restricts the inlet further so it doesn't fall into this category, whereas the Swiftech MCW-462U does), then 3/8" tubing can be better as it offers a faster jet of water onto the hottest sections of the block.

For pretty much everything else, 1/2" is better as it offers less restrictions leading to higher flow rates and hence cooling performance.

[BrianW]

Hey Cathar, quick question: I am planning on getting your block. I was planning on going from rads to the block w/ 1/2" id hose, then after the cpu block, instead of reconnecting the two outlet tubes, I was going to run one tube to the gpu, then the second to the north bridge. Do you think it would be best to use 1/2" id after the cpu block, or reduce down to 3/8"?

Thanks

Brian W

[BillA]

no Cathar
if higher velocity is desired at the 462's inlet, use a fitting having the desired bore at the inlet only
- larger tubing will always be beneficial, but with low flow rates won't make much difference

graphs here: head loss in pipe/hose???

[Cathar]

I'm aware of that Bill, but modifications to the block notwithstanding, a 3/8" central inlet block with be better than a 1/2" inlet block.

Of course, if the block has removable barbs and is 1/2" and has a wide-open 1/2" barb, then rip out the low-restriction barb and go buy a 1/2" barb with an 8mm ID and stick it on. Then basically we have what you're describing Bill (as you've demonstrated), and agreed, that is the better solution.

Problem is many people don't like modifying their blocks so I was answering from that perspective only.

[Cathar]

Originally posted by BrianW
Hey Cathar, quick question: I am planning on getting your block. I was planning on going from rads to the block w/ 1/2" id hose, then after the cpu block, instead of reconnecting the two outlet tubes, I was going to run one tube to the gpu, then the second to the north bridge. Do you think it would be best to use 1/2" id after the cpu block, or reduce down to 3/8"?

You can do what you're describing if you wish. I can even cater for it for you, although I personally feel that the blocks in series afterwards is a better approach. How did you intend do join the two 3/8" outlets from the GPU/NB back into a 1/2" again?

However, depending on what you want, I have no problem with fitting a 1/2" inlet, and two 3/8" outlets for you. My only real concern is that if there is a large pressure-drop imbalance between the NB and GPU blocks (as there probably will be), then the flow within the CPU block will be imbalanced resulting in reduced cooling performance.

If, however, you already own these 3/8" barbed blocks and they cannot be changed to 1/2" easily, then doing what you're suggesting may be the best choice you have open to you, and getting a 3/8" to 1/2" Y joiner from somewhere like http://www.mcmaster.com/ will solve the joining problem. I won't be able to supply that Y myself though for you as I don't live in the USA and source a single one to Australia could be somewhat costly.

[Cathar]

Sorry for the thread hijack, but some 3/8" to 1/2" Y joiners can be obtained here:

http://www.qosina.com/catalog/part.asp?partno=60131

[BrianW]

Thanks Cathar.

I have yet to get the GPU or the NB block, but i was planning on using the new ones from www.dangerden.com. The pressre drop in those two blocks seems similiar to eachother. They can also be purchased in 1/2", but I was figuring that if i use 3/8" the flow through the NB and GPU would be a bit faster, and the coolant would be more effective. What are your thoughts? A second thought: Since the barbs are prolly 1/8" or 1/4" NPT then the water velocity would increase through the blocks regardless if I use 3/8" or 1/2" tube, so it prolly makes sense to go 1/2" the whole way. I can go either way, just would rather get some inout from the masters.

Brian W

[Cathar]

A smaller inlet does not boost water velocity through the block, but it will boost the water velocity where the water initially enters the block. As a result of splitting the flows and using the small inlets, you would probably see decreased cooling performance as a result.

IMO, the best way to do this is:

pump -> rad -> wb -> gpu -> nb -> (res - optional) -> pump

All in-line, all 1/2". This gives the highest water velocity throughout each element in the system. As long as the GPU and NB are low pressure drop blocks, you should not see flow rates suffer much at all, and in fact, maybe no less than if you did the separate 3/8" split approach with the extra restrictions that 3/8" fittings provide.

[BrianW]

Sounds good Cathar, thanks for ur input.

Brian W

[Node]

Cathar, I'm highly considering your block for my system.

You have any favorite waterblocks for gpu and NB?

I'm also considering using two 80mm Black Ice Micro radiators. But I have to decide to run them in parallel (one to waterblock then to northbridge, and the other to GPU and then to vram and HDD) or just run in series through both and then to cpu, then gpu, then northbridge if i do run through the northbridge at all.

[myv65]

Originally posted by Cathar
A smaller inlet does not boost water velocity through the block, but it will boost the water velocity where the water initially enters the block. As a result of splitting the flows and using the small inlets, you would probably see decreased cooling performance as a result.

IMO, the best way to do this is:

pump -> rad -> wb -> gpu -> nb -> (res - optional) -> pump

All in-line, all 1/2". This gives the highest water velocity throughout each element in the system. As long as the GPU and NB are low pressure drop blocks, you should not see flow rates suffer much at all, and in fact, maybe no less than if you did the separate 3/8" split approach with the extra restrictions that 3/8" fittings provide.

Brian,

Understand the trade-off. Cathar's approach will give the highest velocity through each component, but a lower overall flow rate. Since the CPU is generally regarded as needing the most effective cooling, placing all in serial may not be the best approach.

The only guarantees for series vs parallel is that velocity in each component will be higher in series, but total flow rate will be higher in parallel. In your scenario all flow goes through the CPU block either way. As such, it'll see higher velocity with your psuedo-parallel of the GPU/NB. If the GPU/NB blocks are comparitively restrictive vs the CPU block (often the case), then paralleling the final two often proves best.

When in doubt, the best advice is always to test both options and stick with the one that is most effective.

[redleader]

Cathar, what is the address of your site?