Sign in with
Sign up | Sign in
Your question
Solved

Poor temps on GTX 560Ti after loop rebuild

Last response: in Overclocking
Share
a b K Overclocking
December 10, 2011 3:11:55 PM

Howdy all,

I've hit a snag in my loop rebuild and since the important info was getting buried in my build log I figured I'd make a new thread to make it easier.

Only thing I changed was the GPU, pump housing and some QDC fittings (see progressive build timeline).

Quote:
Loop before rebuild:
MCP655 + Koolance housing
360 + 120 rad
i5-2500K + EK Supreme HF Full Copper
2x6950 2GB + EK 6970 Nickel/Plexi Full Cover Blocks

Temps: 45C on both GPUs at load



Quote:
Rebuild Attempt #1
MCP655 + Koolance housing
MSI GTX 560Ti Hawk + Swiftech MCW82
Koolance V4N QDCs by GPU (to facilitate upgrades)

Pump got air in it and tended to "scream," so I ordered a Primochill Myriad dual bay res/top. Never got the computer running long enough to do anything with it (check temps, etc)


Quote:
Rebuild Attempt #2
MCP655 + Primochill Myriad
QDCs still in the loop

Temps were at ~42C idle, and would spike or drastically increase with small uses of the GPU (literally just my changing background every 10 minutes). Eventually the temps would stabilize at 62C. Tried to test with Skyrim and computer shutdown due to overheating within 30 seconds of running the game.


Quote:
Rebuild Attempt #3
Reseated the GPU Block, used MX-2 instead of Ceramique 2 TIM

Absolutely nothing changed from Attempt #2. Absolutely nothing...


Quote:
Rebuild Attempt #3
Took out the QDCs (postulated to block the flow in loop)

Initial idle temps dropped to 38C, but same thermal spiking occurred and eventually leveled off again at 62C. Tested using Furmark to see what the temps were doing, and the temps spiked to 90C+ in under 5 seconds.

More about : poor temps gtx 560ti loop rebuild

a b K Overclocking
December 10, 2011 3:14:17 PM

I'm just completely stumped as to what is causing the problem. I took out all of the variables I could have, and nothing has improved. It's as if there's absolutely no cooling going on in the loop whatsoever. I have TONS of rad space, so I feel like I shouldn't ever be hitting 50C at the most.

My only thoughts now are that the pump isn't moving anything (which I'm sure it is, as I can see tiny bubbles shooting around in the bayres), or that the block is just not transferring heat at all.
m
0
l
a c 331 K Overclocking
December 11, 2011 1:34:47 AM

Have you run the pump with the stock housing to see what it does?

Are you sure you don't have an obstruction inside a block? You removed the QDC's, right? (Pretty sure you said you did). I don't think thermal paste is your problem at all. I'll need to take a look at your loop...how do you currently have it setup?



This is a weird one...if you need quicker responses, I'm on Skype- PM me if you need anything.
m
0
l
Related resources
a b K Overclocking
December 11, 2011 1:46:14 AM

I'm considering breaking it down to the basics and testing the pump. I checked the inside of the GPU block too and no problems. There is water flowing, so I'm wondering if I just have the mother of all air bubbles in one or both of my rads.
m
0
l

Best solution

a c 331 K Overclocking
December 11, 2011 2:06:29 AM

Well, that's a possibility. If you can at least pull the pump and test with some short pieces of tubing in a bucket, etc, you can easily tell if it is working. I will usually use some leftover tubing and use a large bowl in the sink, jumper an extra PSU I have to power the pump and then simply run the kitchen faucet into the bowl to maintain enough water once I start the pump...just make sure you keep the outlet hose under control- some pumps can really push some water (considering we have the same pump, I know the D5's can shoot across your kitchen if you aren't careful).
Share
a b K Overclocking
December 11, 2011 2:24:39 AM

I have a 5 gallon bucket I plan to use for that :) 
m
0
l
a c 331 K Overclocking
December 11, 2011 2:45:52 AM

Good call- you should be able to tell pretty quickly, so shouldn't need to run it that long. Just make sure you get the pump primed so it doesn't run dry. (I know that you know- just reminding you....)
m
0
l
a b K Overclocking
December 11, 2011 3:48:10 AM

Ok pump seems good. It was accidentally run dry a bit because it couldn't generate enough suction at one point and the water evacuated the housing. Looks ok minus a small paint chip on the impeller though.

Commencing rebuild and very thorough air removal.
m
0
l
a b K Overclocking
December 11, 2011 4:54:54 AM

Found out that the Myriad reservoir is causing a myriad of problems. First off, there was a HUGE air bubble in the loop (basically the entire 120 rad wasn't filling).

Then I discovered the second problem: how the flow moves. I have the Myriad hooked up wrong, because the pump draws water in, but I blocked the outlet with a plug because it didn't seem like one of the outlets (yes, there are TWO potential outlets, so I lose flow one way or another. Seriously, why would they design something like this?). I had to look up a complete teardown review because there was no instruction manual to explain all this, and no way to safely disassemble the reservoir.

So in the end, I had absolutely NO water moving through my loop. At all. Ever. I was lucky I didn't blow my pump or one of the plugs.

Going to try again, but may need to go grab some more distilled water...thank god I live in a college town where things are open until god-knows-when. Can't confirm this is exactly the solution, but right now it's looking like it.
m
0
l
a b K Overclocking
December 11, 2011 5:24:55 AM

Just saw the most beautiful thing when I hooked it up again: idle temp 20C :D 

Running Furmark and I'm hitting about 32C steady state and 35C max - but it is cool here now that it's winter.
Now I'm debating putting those QDCs in again, since they don't seem to be the problem...

Also, I would advise staying away from the Myriad unless you want to run a parallel setup on a single pump. There's water coming out from a plate because of the plug fittings. I've linked a thread to the res teardown so it hopefully makes some kind of sense.

I hope this picture explains the flow properly. If not, you're just as confused as I was, but with more available information.



http://forums.bit-tech.net/showthread.php?t=218819
m
0
l
a b K Overclocking
December 11, 2011 5:25:32 AM

Best answer selected by boiler1990.
m
0
l
a c 331 K Overclocking
December 11, 2011 1:57:47 PM

Well, I'll be damned...

That seems like a total PITA, but glad you got it sorted out. Always a good thing when you learn some of those 'gotcha's about your build and are able to adjust accordingly. I was getting worried that you had a bum pump or something.
m
0
l
a b K Overclocking
December 11, 2011 2:06:52 PM

I knew it was running, but I just couldn't figure out where :/  It actually runs dry pretty decently (I know it's not good for it) so maybe I've got a really good one :)  I'm lucky I didn't spring a leak given how powerful the pump is. It's quite annoying that I need a parallel connection (like a Y or something) to fully use my pump's capacity.

I also figured out that the old Koolance pump base should have been mounted vertically (so you could actually fill the reservoir, and so that the inlet and reservoir outlet went down into the pump inlet. It makes so much sense now.

I never would have figured this out if you hadn't nagged me to pull out the pump ;) 
m
0
l
a c 331 K Overclocking
December 11, 2011 6:22:50 PM

I wondered if the Koolance res should have been vertically, but wasn't sure if it was needed or not. But, given that the flows were that stagnant, it had to have been something blocking flow or there being no flow at all. However, as you mentioned there bubbles being produced out of the pump, so you knew it had to be attempting to do...something. :) 

In all seriousness though, very happy you got it sorted out. But at the same time, you also learned a lot about your loop components and determined the the stubborn personality that is apparently designed into D5 pumps from manufacturer assembly plans.
m
0
l
!