High temps - opinions needed

[brammers]

Hi all,

I'm concerned I haven't put something together correctly, but before I tear it down I'd like some opinions/advice...

Here's my system specs:

i7-2600k (4.9G, 1.4v)
KFA2 GTX580
Asus P67 deluxe.

Cooling:

Swiftech Apogee XT
EK FC580 + Backplate
2 x PA120.3, Yate Loons @ ~1100rpm
2 x 18w DDC (series)
Single loop in radbox.

Radbox side of the CPC QDCs is 7/16 Tygon (cloudy as hell now )
PC side of QDC is 3/8 PrimoChill.

Temps:

Ambient is around 22C.

I'm reasonably OK with my load temps of ~74C on the CPU cores when Prime95 is running in conjunction with furmark.

My concern is the 580 ... it's loading at ~85C! This can't be right on water - can it?

I've used the EK standoffs on the 580 board so I didn't warp it and used MX-2 on the 580 chip itself with the supplied pads elsewhere for TIM.

So here's the thing ... did I make a bad mount on the 580, or have I just found the thermal limit of the kit on 1100rpm yates? Upping the fans to 1400rpm does bring temps down by 1 or 2 degrees.

All insights welcome before I strip down and remount the FC580.

TIA
J

[Sideroxylon]

You've definitely got a bad mount on that 580. Mine loads at about 45c.

[brammers]

!

How many mounts can the thermal pads take before they need replacing? Half a dozen or so if I'm careful?

J

[skinnee]

The EK pads are quite durable, 3-6 mounts I would say.

[tiborrr]

I'm reusing them all the time. If you peel them gently they should last for ages, especially 1mm- and thicker.

BTW: You've got a bad mount on that 580GTX.

[brammers]

Thanks guys.

Guess I know what I'm doing tomorrow then

J

[brammers]

Remounted the FC580, doesn't seemed to have helped as much as I'd hoped.

Doubled checked all the pads made good contact.

The MX-2 was spread out nicely over the core - so the mount couldn't have been as bad as I had expected.

Re-applied the MX-2 spreading it as a thin layer this time and remounted - made sure all the screws were good and tight (but I didn't torq them!).

There doesn't seem to be any air in the system.

Furmark is back in the 80's C again (no prime95 running at this point).

PS: Is the temp reporting a probe inside the GFX core, or mounted on the board somewhere? Wonder if the backplate is affecting the numbers.

Help?!

TIA
J

[Toolius]

maybe just try and tighten the screws a little more ?
id say half or 1 rotation only per screw and check ?

[spitter3]

Is the water hot??? Water hot/blocks hot a rad issue, air pocket in the rad or a fan issue. Cold water/hot blocks- flow issue. Cold water/warm blocks-mount issue

[brammers]

The water does get warm yes.

The copper of the EK block get very hot around the 580 core chip. Almost too hot to touch.

I've got two DDC 3.2s in series, so I'd be surprised if it turned out to be a flow issue.

Going to monitor whilst gaming and see what that does, but I'm still not completely happy.

(Kind of hoping EK-Eddy steps in and takes a guess )

[Sideroxylon]

This is really starting to sound like a flow issue. I'd check to make sure your cpu block isn't plugged.

[brammers]

I'll do that when I swap in the rev B3 motherboard.

Might do a full flush and re-fill at that point.

Mind you, the bubbles from the QDC re-joining didn't half move along the tubes when the pumps came on, so I definitely have flow - I guess the question is how much?

I'm going to monitor it during some regular gaming and see if that give any insights as well.

J

[Sideroxylon]

Considering you say you have good contact on the gpu block and that the block is getting hot pretty much narrows it down to some type of flow issue.

What fittings are you using on the gpu block? If the threads are too long they'll hamper flow. I dunno. Just throwing ideas at you at this point.

[brammers]

and I appreciate them!

I'm using a couple of EK 3/8 compressions. They didn't look like they needed the spacers.

Is flow direction important for the EK blocks (I don't think it is looking at it)?

Need to hook up my temp probe in the fluid as well, see what that says.

J

[Eddy_EK]

Send picture of the water block mount imprint to EK support and we'll check it.

[theseeker]

The water does get warm yes.

The copper of the EK block get very hot around the 580 core chip. Almost too hot to touch.

I've got two DDC 3.2s in series, so I'd be surprised if it turned out to be a flow issue.

Going to monitor whilst gaming and see what that does, but I'm still not completely happy.

(Kind of hoping EK-Eddy steps in and takes a guess )

The block should not be hot. I am guessing you have a flow problem too.

[PatRaceTin]

580 on water will not over 45C for that ambient

[brammers]

I'll sort photos out, thanks Eddy.

I'm going to put he 580 back on air too and test, just to make sure it isn't a dud 580 which pumps out silly heat or something like that.

Whilst it's in bits I'll clean out the Apogee XT and flush the whole system (although, blocked in a week? Possible I guess).

You know what, hang it, I'll do a full tear down and clean the tubes etc.

Thank for the help guys - I'll keep you updated.

*yay* back to air for a while

J

[brammers]

So the 580 is back on air.

Ran furmark, temps went straight to 97C as expected (from other online reviews).

Seems as though the card itself is OK.

Looking closely at the EK mount - does flow direction matter at all?

I'm also not convinced that the EK compressions aren't too long in the thread to NOT need the G1/4 spacers for maximum flow - the threads look just a *little* long for comfort.

Anyway, pictures of the mount are below (EK support linked here for reference).

*** 56k warning! they are 8MP files ***

http://www.bramleys.org/radbox/gtx580/

J

[NaeKuh]

thats a OK imprint.

You got good contact.

1. the gpu block should never get hot, unless you have a flow problem.
2. with the amount of radiator you have, i dont think your gpu's should even load at 45.

Last thing i want you to check is for you to make sure you dont have your pump outlet tied to another pump outlet, so water isnt really moving inside your loop from 2 pumps trying to wash in the opposite direction?

Yeah i know its a funny thing to ask, but you'd be suprised... people even forget to plug in there pumps.

[brammers]

No worries, I've done my fair share of dumb **** in my time.

Anyway, here's a picture of the radbox internals, showing the DDC pipework:



All is OK, I'm sure you'll agree.

I'll wait to see what EK say - you never know

Otherwise I'll try another mount / tubing run tomorrow (got a few ideas I want to try).

J

[NaeKuh]

lol u managed to lose me even now.

i dont see what could be possibly wrong besides the card now.

I dont see how the block would even get hot to the touch with that much radiatorage.

Do you see water movement in your res? so you see water being run though the system?

[brammers]

The water is moving.

The small air bubble you get when reconnecting the QDC shifted along very quickly when it powered up

I'm at a loss as well.

I've never seen a report of a dud 580 which overheats - have you?

Oh well, I'll do some information gathering for a bit and see what happens.

J

[NaeKuh]

The water is moving.

The small air bubble you get when reconnecting the QDC shifted along very quickly when it powered up

I'm at a loss as well.

I've never seen a report of a dud 580 which overheats - have you?

Oh well, I'll do some information gathering for a bit and see what happens.

J

thats not the physics buster.

The physics buster is the BLOCK itself should not get hot from all that.
You are reporting the block is getting hot, so that means heat transfer inside the block isnt working.

That tells me either the block must be coated internally with something which is hindering thermal transfer, or I dont know... your somehow breaking physics here...

Basically why is all the heat staying @ your gpu block, and not liking to get disapated at the radiator.

The only way that would be possible is if your rads were pulling in HOT air, so the GPU block was actually cooler because your computer surroundings were colder.

[Sideroxylon]

Yeah contact in those pics look fine.

I almost guarantee you have some type of flow problem. I can't imagine you somehow blocking flow of the gpu block unless it is the threads of the fittings.

Did you open your cpu block to make sure it wasn't plugged up?