not so Quick of a flow question.

[BrianW]

OK

If you have 3/8 hose and you split flow w/ same id connector as the tubing(like the "T" from www.liquidcool.org) and immediately reduce one down to 1/4 id w/ another liquidcool connector and leave the other run the same size 3/8 id, what would be the flow through each tube if the flow in the line before the "T" is X?

Additionally which way of the following would facilitate the fastest flow through A DD Maze II w/ 3/8 id barbs on it, and (2) chipset/video blocks from www.overclock-watercool.com
(they accept 1/4 id tubing). Pump is magdrive 3 350. Rad is black ice. Connectors will all be high flow connectors from www.liquidcool.org .

1) All inline: Res.Pump.Rad.MazeII.NB.Video.Res
All hose will be 3/8 id w/ 14 id,3/8 od slipped over the small blocks to fit the 3/8 id ones over it.

2) All inline: Res.Pump.Rad.MazeII.NB.Video.Res

All hose up to the northbridge will be 3/8 id then will reduce to 1/4 id to fit on the nb and vid. Will reenter the res as 1/4 id tubing.

3) Split right before the cpu like the following:
(3/8 id) Res.Pump.Rad."T" to flow a and b.

a)Stays 3/8 id as it goes to the mazeII then reenters the res independently of flow b.

b)Reduces down to 1/4 id tubing and goes in series from video to nb then back in the res. Again independently of flow a.

4)Split right before the cpu like the following:
(3/8 id) Res.Pump.Rad."T" to flow a and b.

a)Stays 3/8 id as it goes to the mazeII then reenters the res independently of flow b.

b)Stays 3/8 tubing but has those 1/4 id 3/8 od tubes slipped over the barbs on the oc0wc little blocks. Flows in series the same order as #3 flow b. olny diff is the 3/8id tubing.

5)Stays 3/8 id and series like the following:
Res.Pump.Rad.MazeII."T" into flow a and b.

a) Stays 3/8 id and returns directly to the res independenly of flow b.

b) Reduces down to 1/4 id tubing and goes in series from video to nb then back in the res. Again independently of flow a.

6) Stays 3/8 id and series like the following:
Res.Pump.Rad.MazeII."T" into flow a and b.

a) Stays 3/8 id and returns directly to the res independenly of flow b.

b)Stays 3/8 tubing but has those 1/4 id 3/8 od tubes slipped over the barbs on the oc0wc little blocks. Flows in series the same order as #5 flow b. olny diff is the 3/8id tubing.


7) Stays 3/8 id and series like the following:
Res.Pump.Rad.MazeII."T" into flow a and b.

Both flows are identical. One going to the NB, the other going to the Video. This option will use 1/4 id to run through the small chipsets and both return independently to the res.

8)Stays 3/8 id and series like the following:
Res.Pump.Rad.MazeII."T" into flow a and b.

Again the flow will be the same as # 7 but will use the 3/8 id tubing with the small piece of 3/8 od, 1/4id over the barbs on the small blocks.

I am leaning toward 8 right now as I post this thread

[GuyBFF]

Man you really though about this. From all the posts it seems guys are having best luck with either:

Splitting at a distribution block, 3/8" to CPU and 1/4" to NB and 1/4" to GPU, all return to res
-Best overall flow although less flow threw CPU (as half of the water will go threw GPU and NB)

Splitting at a Y, 3/8 to CPU and 1/4" to GPU then NB
-More flow threw CPU (as only competing with a 50% loss), either GPU or NB will run hotter with secondary water

3/8 to CPU then split to 2 1/4" lines for GPU and NB
-100% threw CPU, worst for GPU and NB as secondary water from CPU (although the restrictions from the NB and GPU could lower overall flow rate of system)

I'd say test them as I've read a lot of threads and it really depends on the flow the system has threw all the blocks. If you had a great pump you could afford to distribute the water three ways and the CPU would still get enough flow (I can't say if 350 is enough, due to the restrictions will have the most impact on flow). With less you will need to consider a Y with CPU or after CPU (although a Y after the CPU could slow the flow as water has only two 1/4" paths, before CPU has 3/8 and one 1/4 path).

Nobody seems to have solid data, test and tell us, you have the most comprehensive list

[BrianW]

Where could one purchase a distribution block?

[BladeRunner]

I'll be using some of these eventually, (pic is link to larger image)



Until then I made some:-



Go to www.parkerhydraulics.co.uk and get a cataloge sent out. I posted the pic from the catalogue as the site has no pics. I also get the daily "Darwin Award" for spending 20mins trying to get my POS scanner to work when I realised I can just take a pic of the page with the camera.....

[LiquidCool]

I'm actually going to be making 2 different manifold designs pretty soon...

I'm just curious how you tighten up those 2 fittings without a spot to put a wrench...

[BladeRunner]

Michael Huck

Like most things, when you make or modify stuff you can't find or buy just exactly what you want or need. I made the small aluminium clamp to hold the brass welding bottle fittings, (pictured standing upright). Once turned down I threaded them and they are screwed in tight also using the clamp and some pliers with araldite on the thread to bond & seal them in.

[BrianW]

Blade Runner thanks. Do they copper ones?

Update:

I have decided to go with 3/8 blocks. How to install them with out holes is my next concern.

Should i just go inline? All blocks will be same size. I could split after mazeII and cool the two other chips seperately and have them reenter resevoir seperately.

Thoughts please.

Thanks.

[Def]

Does the initial tubing have to be 3/8" ID?

Why not go 1/2" ID from the pump to the first T, then go 3/8" to the CPU and GPU/NB(in series). Then T it all back up to a 1/2" tube going to the radiator then pump again.

If you have 3/8" fittings on the radiator, and no way to change them, then I'd go 3/8" ID to 3/8" ID for CPU and 1/4" ID for GPU/NB.

Sounds like you really need at least 1/2" ID tubing to really supply that many blocks with a decent amount of water. 3/8" as your biggest pipe just isn't going to cut it.

[BrianW]

3 blocks. CPU,GPU,Video. All will have 3/8 NPT Female threads to accept the barbs. The CPU will have 1/2" barb to accept 12" ID tubing. The other blocks will have 3/8" barbs to accept 3/8" ID tubing.

I will be using two Distribution blocks to get the water to the chips. Here's a pic.

<img border="0" src="http://thengc.net/images/watercooled/guide.jpg" alt="QuickView">

The block on the right will attach directly to the Pump/Res/BleedLine mod I made. Similiar to one on here earlier.

All tubing and fittings will be 1/2" id except for the individual runs to the two smaller blocks. The max head is 24". I have a full tower case, and the rad is in the top and the pump is on the bottom.

[Def]

Ohhh!!! I'm impressed. The efficiency freak part of me is going crazy.

Looks to be a very elegant solution. Although I don't think you will have to really worry about soaking up heat into the "cool" water side. The difference in temp between the "warm" and "cool" sides will only be about a degree. Maybe even less. Keep in mind how much heat it takes to change the temp of water, so you probably aren't going to be dealing with a very large temperature gradient.

[BrianW]

Thanks. The red barb is actually going to be gone as the dist. block will attach to a Res/Fill point, that I just bought at Home Depot for a total of $20. Thanks about the design. I'm a lil proud. Pretty easy with 3d Studio Max.

[BladeRunner]

I just added another outlet to mine for the mobo chipset block. I would have made the manifold longer but that is about as long as it can be so that I can drill a hole straight down without it turning into a nightmare.

As said before it's only temporary so.... I also haven't noticed any rise in CPU temp comparing the system with manifold now, and before when the input tube was just going straight through the CPU block, Maybe it's something to do with the temp of my coolant, (14°C), but I do think people generally get carried away with oversized pipe union Barbs and enormous flow rates. Flow seems more than adequate using a Ehiem 1250 pump which is pumping about 1.5 meters up.

[LiquidCool]

DAMNIT! you keep "taking" my ideas I had been planning this for about 6-8 months, just didnt have the resources to put it all together. Now that I do, someone else does too!

HEHEH Anyway, that is the same thing I had been thinking of. Originally the same as yours, but now I'm going to use a large high flow pump for the runs to the blocks, and a smaller low flow pump for the radiator. All pumping from the same reservoir...

But you do some damn nice work bladerunner. I've seen your other stuff here and at hardwarecentral, very impressive...

[BladeRunner]

Thanks Michael, and I find someone always beats you to things

I'm just about to finally bury this but I just couldn't resist leaving it near the front of my drive all day, It's been getting some very weird looks, but luckily no bomb squad yet

[WaterPog]

I love the bomb, good idea on using the different color hoses...hard to tell once it is in the ground which one is on the top!!

What type of tubing are you using, are you sure that it is "ground contact" rated? I am sure that you thought of this, but it sure looks like air line hose for a compressor system.

[BladeRunner]

Hey you never know, one day I may be able to post a reply without using Pics, but until then ....... The pipes are Polyurethane pneumatic rated at 10 bar and should last for a very long time. (pics are links to detailed image)