ultra low flow loop

[djustice]

I have been setting up my water coolng the past couple of days
the parts are:

-ek full cover WB for 5970 with nickel backpate
-ek mosfet WB
-ek full cover x58 motherboard WB
(all above in copper acryl)
-swiftect micro res
-swiftech xt cu block
-mist 120.2 radiator
-mist 120.3 radiator
-swiftech mcp 655 pump
-1/2" I.D tubing
all fittings are EK high flow 1/2" straigth.
fesser one (red)
flow order: res - pump - 120.3 rad - 120.3 rad -cpu -gpu - mosfet 1- mosfet 2 -motherboard block- res

when i filled the res and started the pump, i could not believe my eyes, the water was moving so fast i could not fill the res fast enough, i had to stop the pump and refill, after filling it as muc as i could, there was enourmous amount of air pockets in the loop.

after juggeling the case upside down sideways, every know directio known to man, i eventully got most of it out.
But when i run the system now it barley trickles water, i also notice about 1/2" difference in the res when i start , shut off pump. there is probably a last bubble trapped in the top radiator.

could this bubble act as a plug? I.E; the pump squeezes water agains the air pocket, compressing it, but not moving it? or have i simply overloaded the pump with too many components?

[luke997]

How do you know that water barely moves, have you got flow meter?

It would be quite weird that suddenly no water flows so perhaps it just looks so as the tubes are full.
Try bleed it some more until you hear no sound when tilting case. With long loop as yours it may take longer for some bubbles to get through all the hurdles (been there

[djustice]

it sure is a pain to get all the air out.

i emptied out the res til the water was below the inlet (but not below the outlet ofc.) now i could see the water puring in theough the 1/2" fitting, and it trickled

EDIT: the feser one fluid seems a bit thicker and stickier then regular distilled water, i guess this is how feser is? hope i have not been shipped bad feser lol

[dejanh]

Better choice for your loop would have been the DDC because it has much higher head pressure. You could very well have reduced the flow that much with all of the components in serial and that pump. Is your D5 set at the highest power setting?

[djustice]

yepp double checked, its at "5" setting

[Petra]

If you're seeing a large drop in reservoir fluid level when you power your pump on, then you have a fair amount of air still in the system--this needs to be removed before you can gauge performance. With a top mounted radiator, some tilting and/or shaking with the pump running will likely be required.

As for the DDC comment, the difference in volumetric output between the DDC and D5 decreases as the total pressure drop of the system increases (read: the difference in flowrate would be minimal). The OP is cooling quite a lot with a single pump, but the D5 should still be able to get the job done (based on curves of similar components, he's probably somewhere right around 1GPM). The OP just needs to get the air out of his system and realize that he isn't going to see a whole ton of activity in that micro res once the res is filled and system bled--observed turbulence in a reservoir isn't useful for determining adequate coolant flowrate.

[wesley]

Sorry to tell you this but this might be hampering your flow & performance in the long run too.

http://www.xtremesystems.org/forums/...d.php?t=226632

Wes

[djustice]

i am finaly up and running, just installed windows, downloading bfbc2 and warcraft.

pc has been idling for some hours, and for this low flow loop i am impressed so far by the idle temps



did a short occt run, it climbed quickly up to 50, but seemd to level out and very slowly climb to 52, after about 3-4 minutes of testing

one run of furmark benchmark did push gpu up to 60 degrees celsius.
but the vrm temps seems as high as ever fushing the 80 mark

[PiLsY]

That Feser premix will gunk up your XT very quickly. I'd strongly suggest you switch to distilled/de-ionised water and either a silver kill coil or couple of drops of Biocide. If you want the coloured look go with coloured tubing . That premix is evil for pin grid blocks - even fin array blocks get gunked given enough time.

Search the forum for the Swiftech GTZ and Feser premix thread - you'll see what I mean . The XT has an even denser pin structure than the GTZ.

You may want to check your gpu mount too. Those temps sound high. Check the vrm pads all have good contact - particularly the smaller ones towards the edge of the board from the main VRM bank.

[Section8]

That Feser premix will gunk up your XT very quickly. I'd strongly suggest you switch to distilled/de-ionised water and either a silver kill coil or couple of drops of Biocide. If you want the coloured look go with coloured tubing . That premix is evil for pin grid blocks - even fin array blocks get gunked given enough time.

Search the forum for the Swiftech GTZ and Feser premix thread - you'll see what I mean . The XT has an even denser pin structure than the GTZ.

You may want to check your gpu mount too. Those temps sound high. Check the vrm pads all have good contact - particularly the smaller ones towards the edge of the board from the main VRM bank.

+1 and +1 to everything Alex said(Petra)

[djustice]

yeah, i have seen those horrible tar like residue deposits screen shots, feser has a high bling factor, that really pulled me in, should have stayed sober when making my desicion on cooling fluid.

anyhow, i have the ati xfx 5970 black edition card (reference design i think), does anyone else have this card with an EK full cover block; i would really like some input as to how many layers of thermal pads they used on the vrm's and what thickness.
(E.g: left side row of vrm's used 0,5 mm +1.0mm pads)

[djustice]

update: i think i found the reason for the low flow, i tore down the rig due to a air venting problem out of my res. (it had bubbled voilently i guess, because i woke up in total darkness, got up and saw that my fuse was blow, check for leakage, could not see any leakae, go out in the stairway, open my fusebox and re engage my fuse. go into my compter room flip the on switch and start the pc; oh my; lol, it sounds like someone is frying bacon wrapped in tinfoil in a micro
now the main fuse blows before i can reach any power plug or fuse box. after a closer inspection, my psu is soked with fesser,i notice that the barb on the top of my res is not completley secured, and cooling liquid had guzzeld up and run down the backside of the ress and straigth in the psu fan and filled up at the bottom of the psu pcb, making contackt with pretty much all solds.

so i orderd a new hx 1000, dont even want to try and start the fesser soked one again.
Now, i tore down the whole ting and drained the lop, good time to flush out fesser one, and use distilled water. i try to flush out the fesser in my swiftech xt block and notice its barley passing trough, now filling the line with distilled water and trying to blow it trough i can barley do it. so i open the block and find the black gooey thing already stuck in the middle of my block. and this is from running the loop for only 14 days!

[Erklat]

Yeah them leaks are pita.
I digg you cause this one time my CPU block has leaked
on the base and block junction, I guess the o-ring wasn't seated properly. The water
slid into the socket producing the same bacon frying sound you described, instantly killing
it even though it was distilled, but it gets conductive over time nevertheless. Luckily it was
an old P35 board which held next to no value then. Though it forced me to upgrade

[Philwong]

So much for the non-conductive property.... man, sorry to hear your loss.

Phil

[WhiteFireDragon]

aww another damage computer due to a leak. i see threads like this once in a while, and it's such an expensive lesson for a simple problem that could have been avoided. like you have found out, colored fluid builds up gunk in the CPU block pins. this also happened to my swiftech block. i was an idiot though, i read the thread on this gunk build up, yet still used dye in my loop

[djustice]

good to hear i am not alone

and indeed its time to change the loop for the better, found these http://www.frozencpu.com/products/88...l=g30c101s1035

miracle workers that i am waiting on i hope to shorten the tubing length due to too thigth turns as it is now, making the tubing flatten.

hopefully within next week i have a lot better layout and moved the res faaaar away from the psu

[djustice]

RE-build:

first off; sorry about the horrible image quality iphone can do a lot of things good, but pictures is not of its strong side, atlest in my untrained hands.


a lot better this time. its currently leak testing with a jump started secoundary psu (not enough to power the pc, but enough for the pump) new psu should be here on tuesday, atlest that what the shipping mail said.








cable management on the the backside of motherboard , inclusve fan regulator and fan power hub



had one tiny leak, 1 drop pr. 10 minutes, was a faulty fesser 45 degree fitting, luckily i had one extra, draind system replaced and now continuing testing, no sign of leaking so far, been 3 hours.

now, does anyone spot any potential trouble with this loop setup? or any improvments that can be done?


Edit: the flow is much better now, i see movment in the res , drained ress to expose inlet barb, and it was flowing good

[Johnny87au]

Looks good to me, Cant find any flaws.. EXcept... Sleeve your cables

[djustice]

yeah, i was thinking about sleeving, i migth go shrink-wrap insted with the same color as the primochill tubing. on the pictures its just the skeleton of the case whith all wires hooked up, but unplugget on the psu side.

still a little work in progress

[Johnny87au]

Never used shrink wrap, Used some of the frozen computer stuff thats pretty good, but my last 2 builds ive been sticking with MDPC sleeving, top grade stuff!