Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   Water Block Design / Construction (http://forums.procooling.com/vbb/forumdisplay.php?f=37)
-   -   Splitting flow from CPU block? (http://forums.procooling.com/vbb/showthread.php?t=9159)

Andersen 03-11-2004 09:20 AM

Splitting flow from CPU block?
 
Newbie here so handle me gently. :)

I'm designing my first set of waterblocks atm. Going to have CPU, GFX and NB blocks.

And the big question is: Is it worth making a single 1/2" inlet and 1/4" outlet CPU block that splits the flow to GFX and NB blocks?

|kbn| 03-11-2004 10:07 AM

No, I think its best to have them all in series, as long as there not too restrictive. If they are restrictive then one will have all the flow while the restrictive one wont get much going through it. woulnt having them in series also need less tubing?
Also I would definatly still use 1/2" outlets as the tubing will still fit over them.

pdf27 03-11-2004 11:03 AM

Depends. If the two (GPU and NB) are about equally restrictive then having the two in paralell will reduce the pressure drop across the system. This will give you increased flow rates and so better cooling.

rundymc 03-12-2004 02:58 AM

also, if the two blocks are not equally restrictive (though they probably will always be slightly so, due to the fact that the tubing on each line will be of different lengths) different amounts of liquid will flow through each- more on the least restrictive- approximated Kirchoff's voltage law
I say design both gpu and nb blocks identically and follow your plan of splitting the flow outside the cpu block (don't use 1/4", as it only adds restriction) with a Y fitting

Andersen 03-12-2004 10:16 AM

Quote:

Originally Posted by rundymc
also, if the two blocks are not equally restrictive (though they probably will always be slightly so, due to the fact that the tubing on each line will be of different lengths) different amounts of liquid will flow through each- more on the least restrictive- approximated Kirchoff's voltage law
I say design both gpu and nb blocks identically and follow your plan of splitting the flow outside the cpu block (don't use 1/4", as it only adds restriction) with a Y fitting

Actually, my designs for NB and GFX block are identical inside. :)

I'll post a render or two later.

Andersen 03-12-2004 10:51 AM

1 Attachment(s)
Quote:

Originally Posted by Andersen
I'll post a render or two later.

As promised. :)

I'm gonna redesign the CPU block so it's not there yet.

|kbn| 03-12-2004 02:29 PM

What components will they be cooling?
If you dont mind some constuctive critasism, i can think of several improvements for them that may even be easyer to make.
Instead of a Z shapped large channel, make several small channels so that you get a lot more surface area and more effective cooling. It would make it more restrictive but not by that much, as you could have them quite wide so it would be almost the same.
Also I would think this way would be easyer to make and possibly waste less material, Ive done similar with my gpu block, and I used a dremmel to cut the channels. what tools do you have to make yours? a drill press or mill would be the only easy way to make those big channels quickly. Or you could do something like a #Rotor by dirlling very close together holes, so that the area left creates pins for hte water to flow over.
Also from the colour in the renders would that be alu your intending to use or copper? If you can get copper I recommend it, as alu will give corrosion problems, worse performance and although its cheaper, I think copper is worth the extra for that.
What program did you use to do the renders btw? they look good :)

BladeRunner 03-12-2004 02:39 PM

I've had a fair bit of experiance with this, :)

Flow splitting can work well especially when you get to three or more blocks in the system.

I've got seven, so my system is a little different, but I'd say a 1/2" cpu block inflow barb and two 3/8" outlets each suppling one of the other blocks also fitted with 3/8" barbs would work fine, and should give less overall flow restriction to the system and most import block.... the CPU.

http://www.zfz.com/pictures/jpg/pcspecs/spec04s.gif

Andersen 03-12-2004 03:01 PM

Quote:

What components will they be cooling?
CPU, GFX (R9700Pro core only) and NB.

Quote:

Instead of a Z shapped large channel, make several small channels so that you get a lot more surface area and more effective cooling. It would make it more restrictive but not by that much, as you could have them quite wide so it would be almost the same.
Yes, that would be better. However, as these would be first blocks I'm gonna make, so I'll keep them simple. ;)

Quote:

Also I would think this way would be easyer to make and possibly waste less material, Ive done similar with my gpu block, and I used a dremmel to cut the channels. what tools do you have to make yours? a drill press or mill would be the only easy way to make those big channels quickly.
I am limited to tools my father currently has. Pillar drill (or is it drill press?), hacksaw, files and crapload of drill bits. :D

I MIGHT get access to a CNC machine but its too uncertain to rely on. Besides, I'm used to make things with my hands. Runs in the family. :D

Quote:

What program did you use to do the renders btw? they look good
3D Studio Max 5.x and Brazil R/S. Not exactly CAD program but I love it. :D

|kbn| 03-12-2004 03:25 PM

3 Attachment(s)
Im also cooling a 9700, but a non pro. With vgpu at 1.75v and overclocking the core to 450mhz, I was limited by the heat it was producing when just using a flat silver base (on the inside, no channels or pins etc...). Ive been chaning the waterblock recently so that It can remove all the heat better. Pic attached if you want to see what done.
As for tools, to do the channels you dont need any expnsive tools. I just used a hand drill and a cutting disk and I think Ive done it well... I could have cut much thinner channels with a different disk but I decided this would be quicker. It took me about an hour or 2 to do this.

first pic is off the channels. in background there is the lid I intended to use, but it was too hard to solder together, so ive made a simpler plastic lid. The drill thing there I used to clean up the channels afterwards.
Pic 2 shows it from the top what the channels/lid will be like.
pic 3 shows how it was before and the cutting disks I have, I used the brown reinforced one for the channels.

have a look at Bladerunners site, this page shows how he made a block that is similar to your design but with different channels. If you make this type of channel, that will probably be the best way to make it without a mill,
http://www.zfz.com/projects.asp?requ...quidgf3&page=4

Bladerunner I noticed you have 90o brass elbows for your gpu block on that page, would you have any spare or know anywhere online I can get some? Also how good was araldite at holding in the barbs?
Your site is good and I learnt a lot from reading it btw.

BladeRunner 03-12-2004 04:56 PM

The brass 90° barbs were made up from two pieces so I could screw them past each other in the block holes. I don't use them any more now as I prefer the Festo style push fittings. the 90° elbows were sourced from a local hydraulics supplier.

Araldite is used purely as a sealing agent, because the threaded barb screws in tightly. I only use the original Araldite, (not Rapid or 5min), as it's much stronger and dries clear, except on my acrylic splitter units where the 5 min stuff seems preferable as it bonds to arylic better. I would never trust any glue of any type, to hold a block together, or barb in, (including silicon sealant).

Gooserider 03-12-2004 09:11 PM

I would point out that it isn't strictly necessary to have the NB and GPU blocks be identical flow resistance - in fact it might be good if they aren't! My understanding is that the NB is relatively low heat output compared to the GPU, so it might be a more effective setup to have the NB block be more restrictive than the GPU block. The NB doesn't need as much cooling, so why waste the coolant on that, when you need it more over on the GPU. Look at BR's setup, note how he has several blocks on the NB branch, but actually uses two branches for the graphics card.

Also BR mentioned this in passing, but I would emphasize more, you need to think more about your tubing sizes. The 1/2" in is good, but the two 1/4" outs would be way restrictive as they have much less area than the 1/2" inlet. You should use 3/8" hose for the outlets as 2 x 3/8" is much closer to 1/2" in volume capacity than 2 x 1/4" Figure the cross section area of each tube and you'll see what I mean! 1/2" = 0.2"^2, 3/8" = 0.11"^2, and 1/4" = 0.05"^2, so you'd need FOUR 1/4" outlets to get the same cross section as a 1/2" tube. It's actually even worse than this since the flow resistance of smaller tubes is greater due to the increased wall surface area.

pdf27 03-13-2004 03:34 AM

Depends where the splitter is. If the splitter is after the block, that is a good idea. If the splitter is part of the block, having an unequal flow resistance will lead to uneven cooling within the block as the NB exit side will be less well cooled.

BladeRunner 03-15-2004 03:26 AM

Yeah if you do the maths the 5/8" inlet and 4 festo outlets add up to the same internal diameter on splitter-x, and there is also the increased wall surface that will make a slight increase in flow restriction.

The reason for splitting is to aid overall system flow. If you piggy back three blocks in series then you will probably create greater overall flow restriction, I don't think it matters if the splitter is combined in the block or after. If you have a sound design then a splitter CPU block can work fine, without running it through more barbs of a separate splitter unit.

Its hard to state anything as a fact because what will work well with system X and Block set X may not work as well with system Y and block set Y, given flow rate Z.... there are just so many specific real world variables. If I had Two blocks only, CPU and VGA, and the VGA was definitely not a flow restrictive design, including the plumbing to and from it, then in series would be best with CPU block first. If it were likely to be flow restrictive I'd favour the extra NB block as well and splitting in or after the CPU block as said, that way the coolant after the CPU block has two paths that should add up to a greater overall flow diameter combined so aiding overall system flow.

Of course it may also make no actual measurable performance difference at all..... :)

If the two blocks after the cpu were poles apart in flow restriction, (and I can't see why they would need to be), then yes it could possibly affect the performance of the CPU block, but a set-up like that is likely to be flow restricting anyway. Like I said considering the heat added to the coolant in one pass is under 1C, assuming sound design and decent flow rates, it would have to be pretty severe, or combined with weak flow rates, to make a difference. Assuming similar block flow designs and tube runs a VGA & Nb block should be very similar. My system has imbalanced flow on the Splitter-X outlets but as the flow rate is very good and the coolant entry to the CPU strikes the base at the centre it shouldn't affect performance at all.

Andersen 03-15-2004 05:07 PM

Hmm...

I have one version with 1/2" inlet right on core and 3/8" outlet on both sides.

And yes, CPU block would act as the splitter in that design.

Some good news! I can get the blocks milled relatively cheaply in local vocational (sp?) school! Smallest mill bit they have is 1mm so I can add some complex designs in my blocks. :D Total cost should stay well under 150 euros. (Hmm, why cant I use euro sign. It shows up as a question mark?)

Vocational schools here in Finland are known to do quality work and can do just about any task one can throw at them.

Oh btw, I might change to metric system so dont wonder. ;)

Im considering adding somekind of fins or a diamond pin matrix as in Swiftec blocks.

Ill post a render or two in couple hours.

Gooserider 03-15-2004 07:10 PM

Quote:

pdf27
Depends where the splitter is. If the splitter is after the block, that is a good idea. If the splitter is part of the block, having an unequal flow resistance will lead to uneven cooling within the block as the NB exit side will be less well cooled.
Perhaps, if the difference was real radical and there was very little flow going to one block. However if both branches had reasonably good flow, I don't think there would be a noticeable effect on the CPU cooling if there were say a 75/25 split in the output flow rates. In an ideal world, both blocks should optimally get about the same flow, as that would imply the maximum possible flow through the CPU block. However if the GPU is not getting enough cooling, it might be helpful to increase the % flow going to it by either throttling the NB block or preferably decreasing the restriction on the GPU block.

Gooserider


All times are GMT -5. The time now is 03:54 PM.

Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...