HELP! T okay over video card?

[TMonte]

I guess I'm being paranoid. I started mocking up my layout tonight and was wondering if there's any danger in running that many junctions over the video card. Not that I'm expecting leaks, and I've never had leak related problems before - so whaddaya think?

Here's a sketch of the setup:



Before anyone jumps in and says the BIP X-flow won't handle the CPU and GPU, just keep in mind I'm not looking for massive overclocks. I'm still considering picking up a PA160 too.

This project is going really well and the case mods are falling together nicely. I'll post some pics when I finish.


EDIT: I also wanted ask if the bottom of the case is better location for the pump - in general, and in consideration of my question above. My initial thought was that up on the drive cage it would have less distance to pump to the Storm, it leaves the PCI slots free (for the most part), and it looks good there...but after looking at it, I wonder if the pump would have trouble pulling the liquid up against gravity. I realize pressure in the loop will help push the liquid along. Any input is very much appreciated.




Thanks!!!
Tom

[gmat]

once your loop is primed the pump location is not important.
Just place everything in order to have the least elbows & turns, the easiest tube routing, and the easiest maintenance. The T location seems good just before the pump - it will help priming it. Just do it

[jman1310]

If you use metal hose clamps and do a proper leak test (<24hrs) it shouldn't be a problem

however

I would recommend putting the T-line after the pump or much further away (<12") so as to not strangle the suction side of the pump. You want the smoothest and least turbulent flow entering the pump. For easiest filling, put it after the pump, in my opinion.

jman

[Mathelo]

Wouldn't it be advisable to have the flow into the "bottom" of the rad and out the "top"? Easier for clearing air in the loop.

L

[TMonte]

Thanks.

@Mathelo: I hadn't thought of it that way. They way I laid it out was under the assupmtion it would be easier for the water to "fall" through the rad rather than have the pump push it up.

Anyone else have thoughts on that?


Tom

[gmat]

air escapes better if it's allowed to go upwards as you'll see the air bubbles will be your worst problem until they're gone...

[TMonte]

I gave it some thought and it's pretty obvious to me now.

I'll run the return from the Maze 4 to the Rad (bottom), and then from the Rad (top) to the pump inlet. The hose bends might be a little trickier, but I guess it'll be easier to bleed.


I'm gonna try to get this done during the week. The plan was to finish up the case mods and assemble over the weekend, but I've been too sick. Not to mention I'm waiting for DD to send along another piece of all-thread I was shorted out of the Maze 4 package (couldn't find 2-56 locally), which should arrive today.


!: edited for clarity.


Thanks,
Tom

[Long Haired Git]

1. Mirroring the pump around and running it as you describe last of all sounds good.
2. Do not put the T in any high pressure area of the flow, so not between pump outlet and the storm, and I'd avoid between the storm and the GPU block. Ideally it goes before the pump inlet, but I suppose between the GPU block and radiator would be best.
3. Putting the pump on the floor of the cage may allow you to run pump -> rad -> cpu block -> gpu block -> T -> pump which means the coolest water gets to the CPU. Sure, only 1/3rd of one degree, but cooler none the less.
4. Re turbulence at pump inlet, I seem to recall most flow is laminar within 5 or so (less than 10, but can't recall exact number) tubing diamaters of the turbulence cause. So, for 10mm ID tubing that's 50mm from the pump inlet. That's for zero effect, so closer than that causes some effect up to a maximum effect when the T would abutt the pump inlet.

[TMonte]

Thanks for the reply.

I always thought putting the rad before the CPU block would be best, but isn't it true that impingement style blocks like the Storm work best with pressure directly from the pump? I saw a DD instructional video where they take a CSP-MAG directly to a TDX. Is that not necessary with a D5?


Tom

[jman1310]

Originally Posted by Long Haired Git

4. Re turbulence at pump inlet, I seem to recall most flow is laminar within 5 or so (less than 10, but can't recall exact number) tubing diamaters of the turbulence cause. So, for 10mm ID tubing that's 50mm from the pump inlet. That's for zero effect, so closer than that causes some effect up to a maximum effect when the T would abutt the pump inlet.

is that ID of the hose diameter or the T-fitting? (just for my background knowledge)

also, is there any concern starving the suction side of the pump?
i remember a post by cather (i think) about using a res directly on the pump to reduce this affect

[Long Haired Git]

Originally Posted by TMonte

Thanks for the reply.
I always thought putting the rad before the CPU block would be best, but isn't it true that impingement style blocks like the Storm work best with pressure directly from the pump? I saw a DD instructional video where they take a CSP-MAG directly to a TDX. Is that not necessary with a D5?
Tom

Not neccessary for ANY block. Pressure degrades from its max at the pump outlet to its min at the pump inlet. Wether it drops a lot at first and then a little at the end or a little at first and a lot at the end is immaterial to the performance of any waterblock.

Putting a pump right on the waterblock may be good due to turbulence with the older designs, but newer designs will not benefit at all.


Next, whatever! Measure the ID as it varies.
Yes, "starvation" of the pump is a factor. Another (better) term for this is "net pressure suction head". Basically, these pumps hate restriction on the inlet. Like 90 degree 1/2" fittings on Eheim 1250's. They perform best when they have a short section of large bore tubing to a reservoir.