SOLVED Curious water cooling problem

[CommanderHL2]

Hi Peoples

I have an existing water cooling setup that has suddenly dropped in performance just with the video card cooling . My GTX480 used to idle at about 50 degrees but has gradually risen up to 70 and is staying there. Still better than the idle temp of 75 with the stock air cooling (I know, its a dog of a video card for heat) but still the cooling should be working a lot better.

The water cooling was obviously working, been that way for a couple of years but something has changed and I don't know what. I took apart my water loop and got rid of the old coolant 3 days ago, used some distilled water to push out the old coolant with my mouth. Put it back together with better joins, 1 less slightly kinked section, some shortened sections and all fresh coolant but the temp is still sitting at 70 idle. CPU is fine, idles at 45-50 depending on the weather, that hasn't changed.

By guesswork I have narrowed it down to a possible radiator blockage or perhaps the pump has weakened. While messing around with the loop from time to time I have started the machine with only the reservoir full of coolant which quickly drains. I switch the computer off as soon as I hear the pump trying to push air, could this damage the pump in just that short moment?

A third possibility is that the video card is faulty and generating more heat somehow but this seems far less likely. The computer is stable and I can still run games for a couple of minutes until the video temp gets over 100 and I have to quit the game.

Any ideas what is the likely culprit here? or ways to find out? I'm looking at buying a flow meter and I have no problem buying a new pump, the one I have now is the MCP355.

Cheers

 

[m0r7if3r]

did you try tightening/loosening the block on the GPU?

 

[Diggrr]

...or removing and reinstalling the block with new thermal goop? It doesn't last forever.

 

[CommanderHL2]

Yeh I took the water block off, removed the old thermal paste and put a fresh spread on. Didn't seem to make any difference.

There is a small bracket which is part of the water block setup which sits around the gpu itself. Its either to prevent thermal paste being pushed onto the card or as a spacer/packer. I was thinking of removing that to see if I can get a tighter seating on the gpu.

 

[robble]

any clogged up parts in the gpu block?

 

[CommanderHL2]

Nah, GPU block is an EK block with a clear cover and there's not many microchannels to start buildup. On the other hand the CPU block is a thermeltake with microchannels i think. That might be worth looking at.

 

[Diggrr]

That sounds like a shim that's meant to protect a nekid core from getting a corner chipped off when you tighten the block down.
If your paste application looks to be thin and even enough when you take it back off, then not much there that can change performance, but to get a tighter mounting you could lap the shim a bit with some 600 grit wet/dry sandpaper.
You can also check to make sure the back plate is still nice and flat, and hasn't bowed over time (weakening the mounting pressure).

 

[CommanderHL2]

Ah good call. The block surface didn't look warped when I took it off but it could be something not easily visible. I'll check that next time I take the block off.

 

[m0r7if3r]

You might be having an issue with the CPU block corroding and causing nastiness if it's a thermaltake. They use aluminum in almost all their products, and it doesn't play well with copper.

 

[CommanderHL2]

This water loop started off as the thermaltake 760i which I was very happy with, pretty sure the cpu block in that kit is copper, its copper on the outside by appearance. I took the drive bay assembly apart to upgrade the pump to the MCP355. I changed the rad to an xspc 240x120. The video card is using an EK nickel/azurite block so all up I've got copper, then nickel, then aluminium rad i think. That might be a bad combo but I havent seen huge amounts of sediment in the reservoir.

The setup might be getting old enough to produce a leak but it would have to be in the cpu block as thats the oldest part. Is copper or aluminium more likely to corrode through first?

 

[CommanderHL2]

Correction, the xspc rad has copper piping, so no ally in my loop thank goodness.

How would I be able to prove that the gpu is warped if this is the case? Put a ruler against the core contact square and look for slight curve? Seems like its going to be tricky to judge.

 

[Diggrr]

Remove it from the video card, hold a razor blade perpendicularly against it and see if you can see light between them (look through towards a bare light bulb for a bright source light).

I was talking about the backing plate on the opposite side of the card from the copper block though. Most seem to be less robust than the copper part of the block, with little thought given to the pressures they must endure. For that you can just use a ruler. Check with the ruler from hole to hole in an X pattern for a 4 screw backer plate.

 

[Modz 4 Higher]

Some good info in this thread. Might I make a suggestion?(not saying its going to fix but I 100% guarantee it will help.) Get rid of that Thermaltake waterblock as soon as budget allows. They are crap.(personal experience) They just need to stick with making narly cases and leave wc'ing to other folks

 

[CommanderHL2]

You might be right about that thermaltake block, modz'. I pulled my loop apart this evening and thoroughly cleaned the rad and cpu waterblock. I did the rad dance maybe 7 times, the 6th time I filled it with vinegar and left it for half an hour, flushed it with distilled water and there was no blue tinge and no gunk. I put vinegar in the cpu block 3 times, 30 mins each, the third time I heated the vinegar to near boiling.

Interestingly I got no blue tinge or gunk from the INSIDE of the cpu block, but some vinegar on the outside turned blue. Can I take this to mean the inside of this cpu block may not be copper? The outside sure as heck is.

Anyhow, the loop is all back together and running, slightly better idle temps, GPU is idling 5 degrees cooler. I fired up a game and the gpu shot up to 100 no probs so no improvement has been made there. Tomorrow I will take the gpu block off check it for warping, been meaning to do it ALL week. I'm predicting its going to turn out to be fine.

My next plan of diagnostic attack will be to buy a new and decent cpu block, **** the thermaltake off, and I'll get a flow meter as the pump is still an area of possible fault.

 

[Modz 4 Higher]

I'm gonna say the TT block is your issue. Do you know the exact model of the block you have? I was gonna try and do some research. I know my TT block was fully compost just a POS. I'll have Ti post up some pics.

Or tell me what is the exact kit you have?

Edit: Did you take the block apart to scrub the inside with a toothbrush? TT blocks tend to have 'micro' channels that get clogged and flushing will not get the crap out. Fortunately for me the one I had was installed with an acrylic top so I could see down inside of the thing easily. This is especially if you used the supplied green uv coolant TT has in their kits. That is the worst crap ever and does not dissipate heat well AT ALL.

 

[CommanderHL2]

The CPU block is from the thermaltake 760i kit, unfortunately this one cant be taken apart. No screws or join, just one solid block. Its a low flow block that has actually worked very well for as long as ive had it, it might be one thing that TT has done right.

I have been using thermaltakes coolant since i started watercooling, havent had an issue with it but as of last night i'm running almost all distilled water. Theres maybe 100ml of coolant still mixed in the distilled water.

 

[CommanderHL2]

YYYEEEEEEESSSS I fixed it!! I got this beast back down to normal temps.

It was a water block/gpu contact issue. Its not a degraded pump or blocked line or warped water block or even a video card issue, thank God. GPU is now idling at 55 with an ambient of 32C, I fired up a game and it maxed out at 83 degrees, slightly better than 100+ I was getting.

There's a shim which comes with the EK gpu water block which I decided not to screw on so tightly in case it was causing a slight warp. I also noticed the supplied standoffs seem to a little too short by maybe ½mm so making those screws really tight was definitely bending the card ever so slightly off the contact surface. I made those screws hand tight and I also pushed the back of the video card core area down on to the water block so the thermal paste was properly spread and sort of glued in place.

I also noticed the water inlet barb thread was too long for the water block. On close inspection there was maybe a 3mm gap for the water to flow into the block. I got the thread ground down by 2-3mm which should have helped the flow quite a bit.

Net result, everything seems to be back to normal, I know what causes this issue and I can start gaming again. I think I learnt much more than I bargained for too.

I took some pics of some of the things I was talking about and I also got a snap of a box cutter blade up against the gpu block to see if there was a warp. If anyones interested I'll post the pics.

 

[Modz 4 Higher]

Pictures are always good and informative

Glad you got your problem/s fixed bro. I know things like this can be very frustrating.

 

[CommanderHL2]

A couple of pics of a box cutter blade up against the core contact area plus a pic showing the standoffs and a pic of the bard thread which I ground down to increase flow into the water block.

 

[Modz 4 Higher]

Very nice pictures man So have you done any test to see if your flow is any better since grinding down the barb fitting? I wonder, if the fitting is not long enough to cover all the threads where it is being mounted if that would make a ripple effect causing turbulence and actually he seeing performance?!?!?! Or am I just thinking to hard